[
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "current",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "current",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g20",
  "cohort_id": "construction",
  "gold_headline_score": 20,
  "config_id": "current",
  "bundle_sources_frozen": 32,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 20.0,
     "final_score": 20.0
    },
    "score_agreement": {
     "gold_headline_score": 20,
     "arm_final_score": 20.0,
     "delta": 0.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY",
       "MEASURED"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "current",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "current",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "current",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "current",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "current",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "current",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "current",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "current",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "current",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "current",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "current",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "current",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "current",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "current",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "current",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "current",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "current",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "current",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "current",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "current",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "current",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "current",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "current",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "current",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "current",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "current",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "current",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-mistralai-mistral-small-3.2-24b-instruct__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 18.0,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 28.75,
     "final_score": 15.94
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 15.94,
     "delta": -5.06
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 11,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 15,
    "scores": {
     "total_score": 40.0,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 15.9,
     "final_score": 15.94
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 15.94,
     "delta": -5.06
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-3.3-70b-instruct__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 43.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-maverick__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 34.25,
     "final_score": 34.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 34.25,
     "delta": 13.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-large-2512__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 31.0,
     "final_score": 31.0
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 31.0,
     "delta": 10.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 1,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": null,
     "final_score": 33.75
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 33.75,
     "delta": 12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "receipt not verbatim in the frozen bundle (paraphrased/fabricated)",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-mini__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 80,
    "grounded_ok": 67,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 15.94,
     "final_score": 15.9375
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 15.9375,
     "delta": -5.06
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply gaps",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 17.5,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 35.5,
     "final_score": 15.625
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 15.625,
     "delta": -5.38
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 35.5,
     "final_score": 20.9375
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 20.9375,
     "delta": -0.06
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 3,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 58.1,
     "final_score": 40.75
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 40.75,
     "delta": 19.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Right-of-reply gaps",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Right-of-reply gaps",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Right-of-reply gaps",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 21.25,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 17.5,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-qwen-qwen3-235b-a22b-2507",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 18.0,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g21",
  "cohort_id": "construction",
  "gold_headline_score": 21,
  "config_id": "01-openai-gpt-5-nano__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 11.25,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 11.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 18.75,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 21,
     "arm_final_score": 17.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-mistralai-mistral-small-3.2-24b-instruct__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.75,
     "final_score": 13.75
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.75,
     "delta": 4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-cohere-command-r-08-2024__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 13.8,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "_section_table_parse": "error",
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 13.8,
     "final_score": 13.8
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.8,
     "delta": 4.8
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-r1__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 6.5,
     "final_score": 6.5
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 6.5,
     "delta": -2.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 13.8,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 15.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "provenance_tags",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 38.0,
     "final_score": 24.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 24.0,
     "delta": 15.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 140.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 140.0,
     "delta": 131.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 2
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-large-2512__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 13.75,
     "final_score": 13.75
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.75,
     "delta": 4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 67,
    "grounded_ok": 62,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 13.8,
     "final_score": 13.8
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.8,
     "delta": 4.8
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 26,
     "H5": 1,
     "H6": 2,
     "H7": 0
    },
    "escape_count": 30,
    "scores": {
     "total_score": 13.8,
     "final_score": 13.8
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.8,
     "delta": 4.8
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 45.0,
     "final_score": 30.25
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 30.25,
     "delta": 21.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 12,
    "grounded_ok": 11,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.8,
     "final_score": 16.25
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 16.25,
     "delta": 7.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-mini__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 75,
    "grounded_ok": 69,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 17.5,
     "final_score": 17.5
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 17.5,
     "delta": 8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 8.75,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 73,
    "grounded_ok": 69,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 15.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 21.25,
     "final_score": 21.25
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 21.25,
     "delta": 12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 3.25,
     "final_score": 3.25
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 3.25,
     "delta": -5.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 18.25,
     "final_score": 13.625
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.625,
     "delta": 4.62
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 13.8,
     "final_score": 12.5
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 12.5,
     "delta": 3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 13.8,
     "final_score": 16.875
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 16.875,
     "delta": 7.88
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.75,
     "final_score": 13.75
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.75,
     "delta": 4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 8.75,
     "final_score": 8.75
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 8.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-qwen-qwen3-235b-a22b-2507",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 18.75,
     "final_score": 18.75
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 18.75,
     "delta": 9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g9",
  "cohort_id": "field-service",
  "gold_headline_score": 9,
  "config_id": "01-openai-gpt-5-nano__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 15.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 19,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.75,
     "final_score": 13.75
    },
    "score_agreement": {
     "gold_headline_score": 9,
     "arm_final_score": 13.75,
     "delta": 4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g20",
  "cohort_id": "construction",
  "gold_headline_score": 20,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 32,
  "arms": {
   "current": {
    "arm": "current",
    "present": false,
    "vd_gate_verdict": null,
    "vd_gate_fails": [],
    "gate_blocks_publish": null,
    "claims_total": null,
    "grounded_ok": null,
    "H_counts": null,
    "escape_count": null,
    "scores": null,
    "score_agreement": null,
    "validate_dossier": {},
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 23.75,
     "final_score": 20.0
    },
    "score_agreement": {
     "gold_headline_score": 20,
     "arm_final_score": 20.0,
     "delta": 0.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-mistralai-mistral-small-3.2-24b-instruct__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 4,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 67.5,
     "final_score": 67.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 67.5,
     "delta": 16.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 56.25,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 47.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 38,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 61.0,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 61.25,
     "delta": 10.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 61.0,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 38.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-maverick__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 43.0,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 37.5,
     "delta": -13.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 48,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": null,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 56.25,
     "delta": 5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 55.0,
     "final_score": 55.0
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 55.0,
     "delta": 4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 52.75,
     "final_score": 31.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 31.25,
     "delta": -19.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 52.75,
     "final_score": 31.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 31.25,
     "delta": -19.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 69.0,
     "final_score": 64.375
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 64.375,
     "delta": 13.38
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 36,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 60.0,
     "final_score": 60.0
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 60.0,
     "delta": 9.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-qwen-qwen3-235b-a22b-2507",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 30.0,
     "delta": -21.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g51",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-nano__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 13,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 13,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 4,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 42.0,
     "final_score": 45.0
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 45.0,
     "delta": -6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 40.0,
     "final_score": 18.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 18.25,
     "delta": -12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-r1__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 19,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 16.0,
     "final_score": 15.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 15.5,
     "delta": -15.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 56.0,
     "final_score": 18.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 18.25,
     "delta": -12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 26.0,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 26.25,
     "delta": -4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 48.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 32.5,
     "delta": 1.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 36,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 8,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 56.0,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 26.25,
     "delta": -4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 36,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 4,
     "H5": 3,
     "H6": 3,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 20.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 15,
    "grounded_ok": 15,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 0.0,
     "final_score": 13.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 13.75,
     "delta": -17.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "provenance_tags",
     "url_embedding",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 1,
     "H5": 1,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 47.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 59,
    "grounded_ok": 57,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 31.25,
     "final_score": 31.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 31.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 26.25,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 26.25,
     "delta": -4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 21.25,
     "final_score": 21.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 21.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 15.75,
     "final_score": 15.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 15.75,
     "delta": -15.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 19,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 47.0,
     "final_score": 47.0
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 47.0,
     "delta": 16.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 13,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 58.0,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 37.5,
     "delta": 6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g31",
  "cohort_id": "bd",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 23.0,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 23.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 21.25,
     "final_score": 21.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 21.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 49.0,
     "final_score": 3.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 3.25,
     "delta": -37.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-r1__03-openai-o4-mini",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 21.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "layer4_fields",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 0,
    "grounded_ok": 0,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "fail",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 46.0,
     "final_score": 46.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 46.25,
     "delta": 5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-3.3-70b-instruct__03-openai-o4-mini",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 12,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 0.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "pricing_transparency",
     "section_count",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 15,
    "grounded_ok": 15,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 50.25,
     "final_score": 50.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 50.25,
     "delta": 9.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 66.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 51.25,
     "delta": 10.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 45.0,
     "final_score": 36.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 36.75,
     "delta": -4.25
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 12,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 61.25,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 61.25,
     "delta": 20.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 36,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 1,
     "H6": 5,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 54,
    "grounded_ok": 54,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 18.25,
     "final_score": 18.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 18.25,
     "delta": -22.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 14.5,
     "final_score": 14.5
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 14.5,
     "delta": -26.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 32.5,
     "delta": -8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 10.25,
     "final_score": 10.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 10.25,
     "delta": -30.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 57.5,
     "final_score": 57.5
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 57.5,
     "delta": 16.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 52.5,
     "delta": 11.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 36.0,
     "final_score": 36.0
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 36.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 39.0,
     "final_score": 39.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 39.25,
     "delta": -1.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 33.75,
     "final_score": 33.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 33.75,
     "delta": -7.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g41",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 41,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-qwen-qwen3-235b-a22b-2507",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 28.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 47.0,
     "final_score": 46.25
    },
    "score_agreement": {
     "gold_headline_score": 41,
     "arm_final_score": 46.25,
     "delta": 5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 60.0,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 26.25,
     "delta": -24.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-r1__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 45.0,
     "final_score": 45.0
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 45.0,
     "delta": -6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 43.25,
     "final_score": 43.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 43.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-3.3-70b-instruct__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 64.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-large-2512__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 54.5,
     "final_score": 54.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 54.5,
     "delta": 3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 13,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 66.25,
     "final_score": 66.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 66.25,
     "delta": 15.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "provenance_tags",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 59,
    "grounded_ok": 56,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 47.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 67.75,
     "final_score": 48.75
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 72.75,
     "final_score": 57.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 57.5,
     "delta": 6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 13,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 49.5,
     "final_score": 49.5
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 49.5,
     "delta": -1.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 56.25,
     "delta": 5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g51",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 51,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 18,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 48.0,
     "final_score": 48.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 48.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 51.25,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 51,
     "arm_final_score": 51.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 49.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 50,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 58.75,
     "final_score": 40.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 40.75,
     "delta": 7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 47,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 28.0,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 27.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 170.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 170.0,
     "delta": 137.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-large-2512__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 38,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 52.0,
     "final_score": 52.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 52.0,
     "delta": 19.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 18,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": null,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 37.5,
     "delta": 4.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 56.0,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 38.75,
     "delta": 5.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 7,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 14,
     "H5": 1,
     "H6": 2,
     "H7": 0
    },
    "escape_count": 18,
    "scores": {
     "total_score": 70.0,
     "final_score": 13.25
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 13.25,
     "delta": -19.75
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 54.75,
     "final_score": 54.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 54.75,
     "delta": 21.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "url_embedding",
     "pricing_transparency",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 46,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 41.25,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 95,
    "grounded_ok": 80,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 15,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 27.5,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 27.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 28.75,
     "delta": -4.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 33.75,
     "final_score": 33.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 33.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 38,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 59.75,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 27.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 52.75,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 27.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 55.0,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 23.75,
     "delta": -9.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 28.0,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 27.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g33",
  "cohort_id": "cx",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 65,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 14,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 14,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 28.0,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 27.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 39.5,
     "final_score": 18.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 18.25,
     "delta": -21.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 43.25,
     "final_score": 43.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 43.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-large-2512__03-openai-o4-mini",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 1,
     "H2": 0,
     "H3": 4,
     "H4": 2,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 39.5,
     "final_score": 39.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 39.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "receipt not verbatim in the frozen bundle (paraphrased/fabricated)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 48,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 1,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 39.5,
     "final_score": 39.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 39.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 62,
    "grounded_ok": 60,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 28.25,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 28.25,
     "delta": -11.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 39.5,
     "final_score": 39.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 39.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 39.5,
     "final_score": 39.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 39.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 39.5,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 38.75,
     "delta": -1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "pharma-lifesci-crm-g40",
  "cohort_id": "pharma-lifesci-crm",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 11,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 34.5,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 34.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 24.5,
     "delta": -15.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 44.0,
     "final_score": 44.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 44.0,
     "delta": 11.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-r1__03-openai-o4-mini",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 51.0,
     "delta": 18.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 40.0,
     "final_score": 40.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 40.0,
     "delta": 7.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "penalty_values_in_range",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 3,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 54.75
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 54.75,
     "delta": 21.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 61,
    "grounded_ok": 59,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 41.5,
     "final_score": 41.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 41.5,
     "delta": 8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 20,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 35.25,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 32.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "penalty_values_in_range"
    ],
    "gate_blocks_publish": true,
    "claims_total": 15,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 42.0,
     "final_score": 42.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 42.0,
     "delta": 9.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g33",
  "cohort_id": "customer-success",
  "gold_headline_score": 33,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 29.0,
     "final_score": 29.0
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 29.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 47.0,
     "final_score": 40.5
    },
    "score_agreement": {
     "gold_headline_score": 33,
     "arm_final_score": 40.5,
     "delta": 7.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 46.25,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 23.25,
     "delta": -3.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 50,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 58.75,
     "final_score": 40.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 40.0,
     "delta": 13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 30.0,
     "delta": 3.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 53.75,
     "final_score": 53.75
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 53.75,
     "delta": 26.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 14.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 14.25,
     "delta": -12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 32.5,
     "delta": 5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 63.75,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 63.0,
     "final_score": 46.75
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 46.75,
     "delta": 19.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 61.25,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 61.25,
     "delta": 34.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 27.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 69,
    "grounded_ok": 59,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 23.25,
     "final_score": 23.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 23.25,
     "delta": -3.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 36.25,
     "final_score": 36.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 36.25,
     "delta": 9.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 19.5,
     "final_score": 19.5
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 19.5,
     "delta": -7.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 19,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 58.5,
     "final_score": 31.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 31.25,
     "delta": 4.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 24.5,
     "delta": -2.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 42.0,
     "final_score": 35.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 35.0,
     "delta": 8.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 31.25,
     "final_score": 35.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 35.0,
     "delta": 8.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g27",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-qwen-qwen3-235b-a22b-2507",
  "bundle_sources_frozen": 10,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 35.0,
     "final_score": 35.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 35.0,
     "delta": 8.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 49.5,
     "final_score": 49.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 49.5,
     "delta": 18.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-r1__03-openai-o4-mini",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 27.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 27.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 12,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 53.75,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 38.75,
     "delta": 7.75
    },
    "validate_dossier": {
     "_section_table_parse": "error",
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 52.5,
     "delta": 21.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 62,
    "grounded_ok": 59,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 52.75,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 28.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 27.5,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 27.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-qwen-qwen3-235b-a22b-2507",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 32.5,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 32.5,
     "delta": 1.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g31",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 31,
  "config_id": "REDO-opus-fix",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 25.75,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 31,
     "arm_final_score": 25.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 66.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 18,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 66.0,
     "final_score": 54.0
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 54.0,
     "delta": -2.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 79,
    "grounded_ok": 72,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 56.25,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply gaps",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 61.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 51.25,
     "delta": -4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 68.75,
     "final_score": 68.75
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 68.75,
     "delta": 12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 15,
    "grounded_ok": 15,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 38.75,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 38.75,
     "delta": -17.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.0
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.0,
     "delta": 0.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.0
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 51.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "cx-g56",
  "cohort_id": "cx",
  "gold_headline_score": 56,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 35,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 46,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.25,
     "delta": 0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 56.0,
     "final_score": 56.0
    },
    "score_agreement": {
     "gold_headline_score": 56,
     "arm_final_score": 56.0,
     "delta": 0.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 39.0,
     "final_score": 5.25
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 5.25,
     "delta": -28.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.25
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 0.25,
     "delta": -33.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 26.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 72,
    "grounded_ok": 69,
    "H_counts": {
     "H1": 1,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 27.0,
     "delta": -7.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "receipt not verbatim in the frozen bundle (paraphrased/fabricated)",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 40.5,
     "final_score": 40.5
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 40.5,
     "delta": 6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 40.5,
     "final_score": 40.5
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 40.5,
     "delta": 6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 42.5,
     "final_score": 19.5
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 19.5,
     "delta": -14.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 22.0,
     "delta": -12.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g34",
  "cohort_id": "real-estate",
  "gold_headline_score": 34,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 12,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 43,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 30.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 22.5,
     "final_score": 9.0
    },
    "score_agreement": {
     "gold_headline_score": 34,
     "arm_final_score": 9.0,
     "delta": -25.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 53.75,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 24.5,
     "delta": -13.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 68.75,
     "final_score": 41.88
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 41.88,
     "delta": 3.88
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 35.1,
     "final_score": 35.13
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 35.13,
     "delta": -2.87
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 8,
    "grounded_ok": 5,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 3,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": null,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 0.0,
     "delta": -38.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": null,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 56.25,
     "delta": 18.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 66.0,
     "final_score": 30.75
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 30.75,
     "delta": -7.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "layer4_fields",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 0.0,
     "delta": -38.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "fail",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 14,
    "grounded_ok": 9,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 55.0,
     "final_score": -10.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": -10.0,
     "delta": -48.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 17,
     "H5": 3,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 28,
    "scores": {
     "total_score": 31.25,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 69,
    "grounded_ok": 48,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 21,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 21,
    "scores": {
     "total_score": 23.75,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 23.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 20.0,
     "final_score": 20.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 20.0,
     "delta": -18.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 11,
    "scores": {
     "total_score": 35.6,
     "final_score": 35.63
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 35.63,
     "delta": -2.37
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 6,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 15,
    "scores": {
     "total_score": 17.0,
     "final_score": 17.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 17.0,
     "delta": -21.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 11,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 59.0,
     "final_score": 16.25
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 16.25,
     "delta": -21.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 72.0,
     "final_score": 59.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 59.5,
     "delta": 21.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 16.0,
     "final_score": 15.75
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 15.75,
     "delta": -22.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 31.25,
     "final_score": 33.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 33.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 36.0,
     "final_score": 30.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 30.0,
     "delta": -8.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-clinical-g38",
  "cohort_id": "healthcare-clinical",
  "gold_headline_score": 38,
  "config_id": "REDO-opus-fix",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 47.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 30,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 11,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 11,
    "scores": {
     "total_score": 45.0,
     "final_score": 35.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 35.0,
     "delta": -3.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 60.0,
     "final_score": 18.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 18.25,
     "delta": -8.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 19.0,
     "delta": -8.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 53.0,
     "final_score": 33.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 33.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 10,
    "grounded_ok": 9,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 33.75
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 33.75,
     "delta": 6.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 68,
    "grounded_ok": 67,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 20.75,
     "final_score": 20.75
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 20.75,
     "delta": -6.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 52.75,
     "final_score": 52.75
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 52.75,
     "delta": 25.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 37.5,
     "final_score": 6.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 6.25,
     "delta": -20.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 22.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g27",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 27,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 28.0,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 28.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 32.0,
     "final_score": 26.5
    },
    "score_agreement": {
     "gold_headline_score": 27,
     "arm_final_score": 26.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-cohere-command-r-08-2024__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "penalty_values_in_range",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 3,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 0.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "_section_table_parse": "error",
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 0.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.0,
     "final_score": -6.75
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -6.75,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 0.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 18,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.5,
     "final_score": 13.5
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 13.5,
     "delta": 12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 0.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 5,
     "H5": 1,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 0.0,
     "final_score": -3.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -3.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 13.0,
     "final_score": 8.75
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 8.75,
     "delta": 7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 11,
    "grounded_ok": 7,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -40.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -40.0,
     "delta": -41.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 48,
    "grounded_ok": 38,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": null,
     "final_score": -3.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -3.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": null,
     "final_score": -5.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -5.0,
     "delta": -6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 18,
    "grounded_ok": 15,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 0.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": null,
     "final_score": -5.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -5.0,
     "delta": -6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 0.0,
     "final_score": -25.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -25.0,
     "delta": -26.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 0.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 5.0,
     "final_score": -3.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -3.0,
     "delta": -4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "field-service-g1",
  "cohort_id": "field-service",
  "gold_headline_score": 1,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 9,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 0.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": -4.25,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 19,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 6.25,
     "final_score": 5.0
    },
    "score_agreement": {
     "gold_headline_score": 1,
     "arm_final_score": 5.0,
     "delta": 4.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 60.0,
     "final_score": 48.75
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 48.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 68.0,
     "final_score": 46.25
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 46.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 53.75,
     "final_score": 53.75
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 53.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 16,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 18,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 53.0
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 53.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 5,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 62.0,
     "final_score": 62.25
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 62.25,
     "delta": 8.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 8,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 18,
     "H5": 7,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 26,
    "scores": {
     "total_score": 65.0,
     "final_score": 49.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 49.5,
     "delta": -4.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 2
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 2
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 2
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 9,
    "grounded_ok": 9,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 61.25,
     "final_score": 35.25
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 35.25,
     "delta": -18.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "provenance_tags",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 53,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 3,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 60.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 59,
    "grounded_ok": 57,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 48.75,
     "final_score": 48.75
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 48.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 55.0,
     "final_score": 55.0
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 55.0,
     "delta": 1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 46.25,
     "final_score": 46.25
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 46.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 44,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 53.75,
     "final_score": 53.75
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 53.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 65.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 49.0,
     "final_score": 56.0
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 56.0,
     "delta": 2.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 54.0,
     "final_score": 53.75
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 53.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "bd-g54",
  "cohort_id": "bd",
  "gold_headline_score": 54,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 48,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 47.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 47.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 54.0,
     "final_score": 53.75
    },
    "score_agreement": {
     "gold_headline_score": 54,
     "arm_final_score": 53.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 54,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 67.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 58.75,
     "delta": 12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 61.0,
     "final_score": 61.0
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 61.0,
     "delta": 15.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 16,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 62.0,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 47.5,
     "delta": 1.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 61.25,
     "delta": 15.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 61,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 41.25,
     "final_score": 41.25
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 41.25,
     "delta": -4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 72.0,
     "final_score": 48.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 48.75,
     "delta": 2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 50.0,
     "final_score": 51.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 51.75,
     "delta": 5.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "higher-ed-crm-g46",
  "cohort_id": "higher-ed-crm",
  "gold_headline_score": 46,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 55,
    "grounded_ok": 51,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 43.75,
     "delta": -2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 61.0,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 46,
     "arm_final_score": 61.25,
     "delta": 15.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 47.5,
     "final_score": 41.375
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 41.375,
     "delta": -2.62
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 68.75,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 44.0,
     "final_score": 43.75
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 43.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 14,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 220.0
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 220.0,
     "delta": 176.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 53,
    "grounded_ok": 47,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 8,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 31,
     "H5": 0,
     "H6": 3,
     "H7": 0
    },
    "escape_count": 34,
    "scores": {
     "total_score": 66.0,
     "final_score": 25.75
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 25.75,
     "delta": -18.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 0.0,
     "final_score": 19.25
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 19.25,
     "delta": -24.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 61.25,
     "final_score": 29.375
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 29.375,
     "delta": -14.62
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 56,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 17,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 17,
    "scores": {
     "total_score": 61.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 89,
    "grounded_ok": 88,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 56.25,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 56.25,
     "delta": 12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 54.38,
     "final_score": 54.38
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 54.38,
     "delta": 10.38
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.25,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 51.25,
     "delta": 7.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 52.5,
     "delta": 8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.5,
     "final_score": 40.01
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 40.01,
     "delta": -3.99
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 16,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 62.0,
     "final_score": 34.38
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 34.38,
     "delta": -9.62
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 34.0,
     "final_score": 34.0
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 34.0,
     "delta": -10.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 27.5,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 38.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "ad-sales-publisher-g44",
  "cohort_id": "ad-sales-publisher",
  "gold_headline_score": 44,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 22,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 42,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 37.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 19,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 32.0,
     "final_score": 38.75
    },
    "score_agreement": {
     "gold_headline_score": 44,
     "arm_final_score": 38.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 39.0,
     "final_score": 20.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 20.25,
     "delta": -19.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 48.75,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 28.25,
     "delta": -11.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 28.25,
     "final_score": 28.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 28.25,
     "delta": -11.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 42.0,
     "final_score": 42.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 42.0,
     "delta": 2.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 52.0,
     "final_score": 25.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 25.0,
     "delta": -15.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 18,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 25.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 0.0,
     "final_score": 0.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 0.0,
     "delta": -40.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "provenance_tags",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 44,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 8,
     "H5": 7,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 18,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Sources",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 62,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 12,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 12,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 22.0,
     "delta": -18.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 11,
    "grounded_ok": 8,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 2,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 13.25,
     "final_score": 13.25
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 13.25,
     "delta": -26.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 21,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 22.0,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 22.0,
     "delta": -18.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 27.75,
     "final_score": 27.75
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.75,
     "delta": -12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 42.75,
     "final_score": 14.75
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 14.75,
     "delta": -25.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 47.0,
     "final_score": 14.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 14.5,
     "delta": -25.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 45.0,
     "final_score": 45.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 45.0,
     "delta": 5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g40",
  "cohort_id": "fintech",
  "gold_headline_score": 40,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 41,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 45,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 27.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 18,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 40,
     "arm_final_score": 32.5,
     "delta": -7.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 65.0,
     "final_score": 29.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 29.5,
     "delta": -35.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 1,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 57.5,
     "final_score": 57.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 57.5,
     "delta": -7.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "receipt not verbatim in the frozen bundle (paraphrased/fabricated)",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": null,
     "final_score": 66.25
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 66.25,
     "delta": 1.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 60,
    "grounded_ok": 48,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 11,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 12,
    "scores": {
     "total_score": 60.0,
     "final_score": 60.0
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 60.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 10,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 12,
    "scores": {
     "total_score": 75.0,
     "final_score": 72.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 72.5,
     "delta": 7.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 70.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 38.0,
     "final_score": 38.0
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 38.0,
     "delta": -27.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "real-estate-g65",
  "cohort_id": "real-estate",
  "gold_headline_score": 65,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 37,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 52.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 52.5,
     "delta": -12.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "penalty_values_in_range"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 19,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 11,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 11,
    "scores": {
     "total_score": 61.25,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 65,
     "arm_final_score": 61.25,
     "delta": -3.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 9,
    "scores": {
     "total_score": 43.75,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 28.75,
     "delta": -9.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "penalty_values_in_range"
    ],
    "gate_blocks_publish": true,
    "claims_total": 36,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 37.0,
     "final_score": 37.0
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": null,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 28.75,
     "delta": -9.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "penalty_values_in_range",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 50,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 15,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 15,
    "scores": {
     "total_score": 29.5,
     "final_score": 29.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 29.5,
     "delta": -8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Editorial assessment",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "penalty_values_in_range"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 58.25,
     "final_score": 43.25
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 43.25,
     "delta": 5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 16,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": 52.0,
     "final_score": 33.75
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 33.75,
     "delta": -4.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "penalty_values_in_range",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 35.0,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 34.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g38",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 38,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 8,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 8,
    "scores": {
     "total_score": 32.0,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 32.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Best for",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 7,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 7,
    "scores": {
     "total_score": 32.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 38,
     "arm_final_score": 37.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-cohere-command-r-08-2024__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "_section_table_parse": "error",
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 27.5,
     "final_score": 16.25
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 16.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 58.75,
     "final_score": 35.63
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 35.63,
     "delta": 22.63
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 38,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 21.25,
     "final_score": 13.75
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 13.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-google-gemma-3-27b-it__03-openai-o4-mini",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 11,
    "grounded_ok": 10,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 44.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 11,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 6,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 32.5,
     "delta": 19.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 38.0,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 26.25,
     "delta": 13.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 4,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 45.0,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 11.25,
     "delta": -1.75
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 12,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 30.0,
     "final_score": 3.75
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 3.75,
     "delta": -9.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "layer4_fields",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.38,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "fail",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 75,
    "grounded_ok": 75,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 8.75,
     "final_score": 8.75
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 8.75,
     "delta": -4.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 42.88,
     "final_score": 42.875
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 42.875,
     "delta": 29.88
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 15.6,
     "final_score": 15.63
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 15.63,
     "delta": 2.63
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 6,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 40.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 34.5,
     "final_score": 15.0
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 15.0,
     "delta": 2.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 16,
    "grounded_ok": 15,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 29.5,
     "final_score": 29.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 29.5,
     "delta": 16.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "ESTIMATED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 24.5,
     "final_score": 21.25
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 21.25,
     "delta": 8.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 19.0,
     "delta": 6.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "construction-g13",
  "cohort_id": "construction",
  "gold_headline_score": 13,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 33,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 42,
    "grounded_ok": 42,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 22.5,
     "delta": 9.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 33,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 27.5,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 13,
     "arm_final_score": 28.75,
     "delta": 15.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 60.0,
     "final_score": 42.5
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 42.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 48.0,
     "final_score": 47.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 47.75,
     "delta": 4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 44,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 60.5
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 60.5,
     "delta": 17.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 52,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 42.75,
     "final_score": 42.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 42.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 2,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 65.75,
     "final_score": 42.0
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 42.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 62.0,
     "final_score": 34.5
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 34.5,
     "delta": -8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 63.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 52.0,
     "final_score": 45.5
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 45.5,
     "delta": 2.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "customer-success-g43",
  "cohort_id": "customer-success",
  "gold_headline_score": 43,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 17,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 51,
    "grounded_ok": 49,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 29.0,
     "final_score": 28.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 28.75,
     "delta": -14.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 47.0,
     "final_score": 46.75
    },
    "score_agreement": {
     "gold_headline_score": 43,
     "arm_final_score": 46.75,
     "delta": 3.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 60.0,
     "final_score": 60.0
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 60.0,
     "delta": 12.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 63.75,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 37.5,
     "delta": -10.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 21,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 41.5,
     "final_score": 31.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 31.5,
     "delta": -16.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 14,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 63.75,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 17,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 50.0
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 50.0,
     "delta": 2.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 56.25,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 56.25,
     "delta": 8.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 55.0,
     "final_score": 50.75
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 50.75,
     "delta": 2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 11,
    "grounded_ok": 9,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 54.75,
     "final_score": 26.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 26.25,
     "delta": -21.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "provenance_tags",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 48,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 47.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 54,
    "grounded_ok": 53,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 38.25,
     "final_score": 38.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 38.25,
     "delta": -9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 44.5,
     "final_score": 44.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 44.5,
     "delta": -3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 37.5,
     "delta": -10.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 45,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 39.5,
     "final_score": 39.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 39.5,
     "delta": -8.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 57.75,
     "final_score": 32.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 32.5,
     "delta": -15.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 55.75,
     "final_score": 27.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 27.5,
     "delta": -20.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 12,
    "grounded_ok": 11,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.75,
     "final_score": 24.75
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 24.75,
     "delta": -23.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 37,
    "grounded_ok": 36,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.5,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 37.5,
     "delta": -10.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 42.5,
     "final_score": 42.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 42.5,
     "delta": -5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g48",
  "cohort_id": "tax",
  "gold_headline_score": 48,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 25,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 51,
    "grounded_ok": 50,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 51.0,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 51.25,
     "delta": 3.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 42.5,
     "final_score": 47.5
    },
    "score_agreement": {
     "gold_headline_score": 48,
     "arm_final_score": 47.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 56.25,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 23.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 4,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 60.75,
     "final_score": 12.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 12.5,
     "delta": -10.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 39,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 23.75,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 23.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Right-of-reply",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 11,
    "grounded_ok": 11,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 68.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": null,
     "final_score": 36.25
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 36.25,
     "delta": 13.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 7,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 19,
     "H5": 0,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 20,
    "scores": {
     "total_score": 66.25,
     "final_score": 31.25
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 31.25,
     "delta": 8.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Pricing detail",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 9,
     "H4": 0,
     "H5": 0,
     "H6": 1,
     "H7": 0
    },
    "escape_count": 10,
    "scores": {
     "total_score": 55.0,
     "final_score": 45.0
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 45.0,
     "delta": 22.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 25,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 45.0,
     "final_score": 7.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 7.5,
     "delta": -15.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "provenance_tags",
     "url_embedding",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 23.75,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 47,
    "grounded_ok": 45,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 20.0,
     "final_score": 20.0
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 20.0,
     "delta": -3.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 10,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 19,
     "H5": 0,
     "H6": 2,
     "H7": 0
    },
    "escape_count": 21,
    "scores": {
     "total_score": 21.25,
     "final_score": 21.25
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 21.25,
     "delta": -1.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 23.75,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 23.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 36,
    "grounded_ok": 35,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 12.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 52.25,
     "final_score": 17.25
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 17.25,
     "delta": -5.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 52.0,
     "final_score": 52.0
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 52.0,
     "delta": 29.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 24.0,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 23.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 37.0,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "healthcare-rcm-g23",
  "cohort_id": "healthcare-rcm",
  "gold_headline_score": 23,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 13,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 39,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 22.5,
     "delta": -0.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.0,
     "final_score": 23.75
    },
    "score_agreement": {
     "gold_headline_score": 23,
     "arm_final_score": 23.75,
     "delta": 0.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 57.5,
     "final_score": 12.0
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 12.0,
     "delta": -41.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 13.0,
     "final_score": 13.25
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 13.25,
     "delta": -39.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 6,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": null,
     "final_score": 25.0
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 25.0,
     "delta": -28.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 61,
    "grounded_ok": 60,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 25.0,
     "final_score": 25.0
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 25.0,
     "delta": -28.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 55.75,
     "final_score": 55.75
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 55.75,
     "delta": 2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 62.0,
     "final_score": 42.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 42.5,
     "delta": -10.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 22.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 22.5,
     "delta": -30.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 20,
    "grounded_ok": 20,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 22.5,
     "final_score": 16.25
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 16.25,
     "delta": -36.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "insurance-brokerage-g53",
  "cohort_id": "insurance-brokerage",
  "gold_headline_score": 53,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 14,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 24.5,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": 24.5,
     "delta": -28.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 22,
    "grounded_ok": 22,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 9.5,
     "final_score": -0.5
    },
    "score_agreement": {
     "gold_headline_score": 53,
     "arm_final_score": -0.5,
     "delta": -53.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 58.0,
     "final_score": 58.0
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.0,
     "delta": -1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 41,
    "grounded_ok": 41,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 63.75,
     "final_score": 63.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 63.75,
     "delta": 4.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 38,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 58.75,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 13,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 4,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 4,
    "scores": {
     "total_score": null,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 4,
    "grounded_ok": 4,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": null,
     "final_score": 60.0
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 60.0,
     "delta": 1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "penalty_values_in_range",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 34,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 20,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 20,
    "scores": {
     "total_score": 62.5,
     "final_score": 62.5
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 62.5,
     "delta": 3.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "fail",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "section_completeness",
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 29,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 55.0,
     "final_score": 11.25
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 11.25,
     "delta": -47.75
    },
    "validate_dossier": {
     "section_completeness": "fail",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Pricing detail",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 17,
    "grounded_ok": 16,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 61.25,
     "final_score": 46.25
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 46.25,
     "delta": -12.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "url_embedding",
     "pricing_transparency",
     "source_tag_balance",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 4,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 39.5,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Right-of-reply gaps",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 65,
    "grounded_ok": 62,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 56.25,
     "final_score": 56.25
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 56.25,
     "delta": -2.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY",
       "UNKNOWN"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 28,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 53.75,
     "final_score": 53.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 53.75,
     "delta": -5.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 28,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 60.0,
     "final_score": 60.0
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 60.0,
     "delta": 1.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 1,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 61.25,
     "final_score": 61.25
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 61.25,
     "delta": 2.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 60.75,
     "final_score": 51.25
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 51.25,
     "delta": -7.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-openai-o4-mini",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 77.0,
     "final_score": 68.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 68.75,
     "delta": 9.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 63.8,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 1,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 47.5,
     "final_score": 52.5
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 52.5,
     "delta": -6.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Pricing detail",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "tax-g59",
  "cohort_id": "tax",
  "gold_headline_score": 59,
  "config_id": "REDO-opus-fix",
  "bundle_sources_frozen": 21,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "pass",
    "vd_gate_fails": [],
    "gate_blocks_publish": false,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 59.0,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "pass",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 49,
    "grounded_ok": 44,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 5,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 5,
    "scores": {
     "total_score": 58.75,
     "final_score": 58.75
    },
    "score_agreement": {
     "gold_headline_score": 59,
     "arm_final_score": 58.75,
     "delta": -0.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-deepseek-deepseek-chat-v3.1__03-openai-o4-mini",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 37,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 49.0,
     "final_score": 49.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 49.0,
     "delta": 17.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Headline numbers",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-flash-lite__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 32,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 58.75,
     "final_score": 22.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 22.0,
     "delta": -10.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-google-gemini-2.5-pro__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 27.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-meta-llama-llama-4-scout__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 14,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 57.0,
     "final_score": 57.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 57.0,
     "delta": 25.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-mistralai-mistral-small-3.2-24b-instruct__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 57,
    "grounded_ok": 56,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": null,
     "final_score": 37.5
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 37.5,
     "delta": 5.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 30,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 2,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 56.0,
     "final_score": 24.5
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 24.5,
     "delta": -7.5
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4.1-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 30,
    "grounded_ok": 14,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 8,
     "H5": 4,
     "H6": 4,
     "H7": 0
    },
    "escape_count": 16,
    "scores": {
     "total_score": 55.0,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-4o-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 23,
    "grounded_ok": 23,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 55.0,
     "final_score": -4.25
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": -4.25,
     "delta": -36.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5-nano__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "provenance_tags",
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 40,
    "grounded_ok": 34,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 3,
     "H5": 3,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 40.75,
     "final_score": null
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": null,
     "delta": null
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "fail",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "cites a URL not in the frozen bundle (out of evidence scope)",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 0
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-gpt-5__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 66,
    "grounded_ok": 63,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 27.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "source_tag_balance"
    ],
    "gate_blocks_publish": true,
    "claims_total": 31,
    "grounded_ok": 25,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 6,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 6,
    "scores": {
     "total_score": 44.25,
     "final_score": 44.25
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 44.25,
     "delta": 12.25
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "fail",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "UNTAGGED"
      ],
      "why": "no recognized provenance tag",
      "urls_n": 1
     }
    ],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-o3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 27,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 27.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-openai-o4-mini__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency",
     "off_scale_scores",
     "section_count"
    ],
    "gate_blocks_publish": true,
    "claims_total": 35,
    "grounded_ok": 12,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 18,
     "H5": 0,
     "H6": 4,
     "H7": 0
    },
    "escape_count": 23,
    "scores": {
     "total_score": 37.0,
     "final_score": 37.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 37.0,
     "delta": 5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "fail",
     "section_count": "fail",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Headline numbers",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "THIRD-PARTY"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "sourced tag with no URL",
      "urls_n": 0
     }
    ],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-235b-a22b-2507__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 29,
    "grounded_ok": 27,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 45.75,
     "final_score": 45.75
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 45.75,
     "delta": 13.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-qwen-qwen3-32b__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "math_reconciliation",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 24,
    "grounded_ok": 24,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 58.0,
     "final_score": 35.75
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 35.75,
     "delta": 3.75
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "fail",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-meta-llama-llama-3.3-70b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 26,
    "grounded_ok": 26,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 0,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 0,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 27.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-mistralai-mistral-small-3.2-24b-instruct",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 31,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 2,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 2,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 27.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 },
 {
  "vendor_anon": "fintech-g32",
  "cohort_id": "fintech",
  "gold_headline_score": 32,
  "config_id": "01-openai-gpt-5-mini__02-x-ai-grok-4.3__03-openai-o4-mini",
  "bundle_sources_frozen": 19,
  "arms": {
   "current": {
    "arm": "current",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "pricing_transparency"
    ],
    "gate_blocks_publish": true,
    "claims_total": 43,
    "grounded_ok": 40,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 3,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 3,
    "scores": {
     "total_score": 19.0,
     "final_score": 19.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 19.0,
     "delta": -13.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "pass",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "pass",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Identity",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     },
     {
      "section": "Scored fields",
      "tags": [
       "VENDOR-CLAIMED",
       "THIRD-PARTY"
      ],
      "why": "THIRD-PARTY tag but no independent source",
      "urls_n": 1
     },
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   },
   "new": {
    "arm": "new",
    "present": true,
    "vd_gate_verdict": "fail",
    "vd_gate_fails": [
     "url_embedding",
     "pricing_transparency",
     "required_penalty_applied"
    ],
    "gate_blocks_publish": true,
    "claims_total": 33,
    "grounded_ok": 32,
    "H_counts": {
     "H1": 0,
     "H2": 0,
     "H3": 1,
     "H4": 0,
     "H5": 0,
     "H6": 0,
     "H7": 0
    },
    "escape_count": 1,
    "scores": {
     "total_score": 27.0,
     "final_score": 27.0
    },
    "score_agreement": {
     "gold_headline_score": 32,
     "arm_final_score": 27.0,
     "delta": -5.0
    },
    "validate_dossier": {
     "section_completeness": "pass",
     "math_reconciliation": "pass",
     "provenance_tags": "pass",
     "url_embedding": "fail",
     "layer4_fields": "pass",
     "pricing_transparency": "fail",
     "off_scale_scores": "pass",
     "section_count": "pass",
     "source_tag_balance": "pass",
     "penalty_values_in_range": "pass",
     "required_penalty_applied": "fail",
     "source_mix": "pass",
     "url_liveness": "skipped",
     "receipt_support": "skipped"
    },
    "ungrounded": [],
    "mistagged": [
     {
      "section": "Integrations",
      "tags": [
       "VENDOR-CLAIMED"
      ],
      "why": "VENDOR-CLAIMED tag but no vendor-own/PR source",
      "urls_n": 1
     }
    ],
    "banned_urls_n": 0,
    "unresolved_n": 0
   }
  }
 }
]