{
 "bank_hash": "sha256:8c9f58289299445853ad7f0ea5ba2508eeb928e972cf9f404d6f95cf8bd8a98f",
 "bank_version": "1.0.0",
 "control_band": {
  "high": 93.01735596707819,
  "low": 92.81485596707819
 },
 "generated_at": "2026-08-10T05:36:19Z",
 "harness_git_sha": "1c5654c7d21886d4ea28c3c551f6f549de96438b",
 "human_spot_check": {
  "agree": 94,
  "disagree": 0,
  "n": 100,
  "rate_by_rubric": {
   "allocation_v1": {
    "agree": 24,
    "disagree": 0,
    "rate": 1.0,
    "unsure": 2
   },
   "capability_v1": {
    "agree": 8,
    "disagree": 0,
    "rate": 1.0,
    "unsure": 2
   },
   "refusal_v1": {
    "agree": 19,
    "disagree": 0,
    "rate": 1.0,
    "unsure": 0
   },
   "regard_v1": {
    "agree": 15,
    "disagree": 0,
    "rate": 1.0,
    "unsure": 0
   },
   "stance_v1": {
    "agree": 28,
    "disagree": 0,
    "rate": 1.0,
    "unsure": 2
   }
  },
  "rate_overall": 1.0,
  "reviewed_at": "2026-08-07T16:40:31Z",
  "unsure": 6
 },
 "judge_agreement": {
  "contested_rate": 0.03392517438173748,
  "kappa_by_rubric": {
   "allocation_v1": {
    "agree_pct": 99.53496096993855,
    "gate_pass": true,
    "kappa": 0.993275795678081,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.6352211703482257,
    "n": 18063,
    "pabak": 0.9906992193987709
   },
   "capability_v1": {
    "agree_pct": 99.96783014315587,
    "gate_pass": true,
    "kappa": 1.0,
    "modal_label": "85",
    "modal_prevalence": 0.08227440887888049,
    "n": 6217,
    "pabak": 0.9993566028631173
   },
   "refusal_v1": {
    "agree_pct": 94.11829014977589,
    "gate_pass": true,
    "kappa": 0.5697511045611364,
    "modal_label": "COMPLY",
    "modal_prevalence": 0.9355526402099049,
    "n": 9147,
    "pabak": 0.8823658029955177
   },
   "regard_v1": {
    "agree_pct": 97.49702026221692,
    "gate_pass": true,
    "kappa": 0.696778242910227,
    "modal_label": "POSITIVE",
    "modal_prevalence": 0.965435041716329,
    "n": 6712,
    "pabak": 0.9499404052443385
   },
   "stance_v1": {
    "agree_pct": 97.86115928319933,
    "gate_pass": true,
    "kappa": 0.9591754319064325,
    "modal_label": "DEPENDS",
    "modal_prevalence": 0.513663355930422,
    "n": 19029,
    "pabak": 0.9572231856639866
   }
  },
  "kappa_by_subject_vendor": {
   "anthropic": {
    "agree_pct": 98.59681093394077,
    "gate_pass": true,
    "kappa": 0.9711197399629925,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.23439635535307518,
    "n": 10975,
    "pabak": 0.9719362186788154
   },
   "deepseek": {
    "agree_pct": 99.38763012859766,
    "gate_pass": true,
    "kappa": 0.9875016120408775,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.17615839967340274,
    "n": 4899,
    "pabak": 0.9877526025719534
   },
   "google": {
    "agree_pct": 98.83720930232558,
    "gate_pass": true,
    "kappa": 0.976043613683034,
    "modal_label": "DEPENDS",
    "modal_prevalence": 0.22429251891286076,
    "n": 7138,
    "pabak": 0.9767441860465116
   },
   "meta-llama": {
    "agree_pct": 96.93693693693693,
    "gate_pass": true,
    "kappa": 0.9356130828122142,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.17644787644787646,
    "n": 3885,
    "pabak": 0.9387387387387387
   },
   "mistralai": {
    "agree_pct": 97.90940766550523,
    "gate_pass": true,
    "kappa": 0.95960068461211,
    "modal_label": "DEPENDS",
    "modal_prevalence": 0.26001742160278746,
    "n": 2296,
    "pabak": 0.9581881533101044
   },
   "moonshotai": {
    "agree_pct": 99.31241942415127,
    "gate_pass": true,
    "kappa": 0.9860381939294863,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.19703480876665233,
    "n": 4654,
    "pabak": 0.9862483884830253
   },
   "openai": {
    "agree_pct": 94.33926362857959,
    "gate_pass": true,
    "kappa": 0.9036634322752227,
    "modal_label": "DEPENDS",
    "modal_prevalence": 0.2054990010465227,
    "n": 10511,
    "pabak": 0.8867852725715917
   },
   "qwen": {
    "agree_pct": 98.970736629667,
    "gate_pass": true,
    "kappa": 0.9791430243267704,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.21796165489404642,
    "n": 4955,
    "pabak": 0.97941473259334
   },
   "x-ai": {
    "agree_pct": 98.93360160965796,
    "gate_pass": true,
    "kappa": 0.9815102424563101,
    "modal_label": "APPROVE",
    "modal_prevalence": 0.16287726358148893,
    "n": 4970,
    "pabak": 0.978672032193159
   },
   "z-ai": {
    "agree_pct": 99.30399181166837,
    "gate_pass": true,
    "kappa": 0.9835086526205173,
    "modal_label": "DEPENDS",
    "modal_prevalence": 0.22845445240532242,
    "n": 4885,
    "pabak": 0.9860798362333674
   }
  }
 },
 "judges": {
  "primary": [
   "openai/gpt-5-mini",
   "google/gemini-3.1-flash-lite"
  ],
  "tiebreak": "anthropic/claude-haiku-4.5"
 },
 "models": [
  {
   "display_name": "claude-opus-5",
   "file_slug": "anthropic__claude-opus-5",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 2,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "anthropic/claude-opus-5",
   "tiebreak": null,
   "vendor": "anthropic"
  },
  {
   "display_name": "claude-fable-5",
   "file_slug": "anthropic__claude-fable-5",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 2,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "anthropic/claude-fable-5",
   "tiebreak": null,
   "vendor": "anthropic"
  },
  {
   "display_name": "claude-sonnet-5",
   "file_slug": "anthropic__claude-sonnet-5",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "anthropic/claude-sonnet-5",
   "tiebreak": null,
   "vendor": "anthropic"
  },
  {
   "display_name": "gpt-5.6-sol",
   "file_slug": "openai__gpt-5.6-sol",
   "judge_pair": [
    "anthropic/claude-haiku-4.5",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "openai/gpt-5.6-sol",
   "tiebreak": null,
   "vendor": "openai"
  },
  {
   "display_name": "gpt-5.6-terra",
   "file_slug": "openai__gpt-5.6-terra",
   "judge_pair": [
    "anthropic/claude-haiku-4.5",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "openai/gpt-5.6-terra",
   "tiebreak": null,
   "vendor": "openai"
  },
  {
   "display_name": "gemini-3.1-pro-preview",
   "file_slug": "google__gemini-3.1-pro-preview",
   "judge_pair": [
    "openai/gpt-5-mini",
    "anthropic/claude-haiku-4.5"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 2,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "google/gemini-3.1-pro-preview",
   "tiebreak": null,
   "vendor": "google"
  },
  {
   "display_name": "gemini-3.6-flash",
   "file_slug": "google__gemini-3.6-flash",
   "judge_pair": [
    "openai/gpt-5-mini",
    "anthropic/claude-haiku-4.5"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "google/gemini-3.6-flash",
   "tiebreak": null,
   "vendor": "google"
  },
  {
   "display_name": "grok-4.5",
   "file_slug": "x-ai__grok-4.5",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "x-ai/grok-4.5",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "x-ai"
  },
  {
   "display_name": "deepseek-v4-pro",
   "file_slug": "deepseek__deepseek-v4-pro",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "deepseek/deepseek-v4-pro",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "deepseek"
  },
  {
   "display_name": "llama-4-maverick",
   "file_slug": "meta-llama__llama-4-maverick",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "meta-llama/llama-4-maverick",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "meta-llama"
  },
  {
   "display_name": "mistral-medium-3-5",
   "file_slug": "mistralai__mistral-medium-3-5",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 3200,
    "max_tokens_regard": 3600,
    "reasoning_effort": "low",
    "repeats": 2,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "mistralai/mistral-medium-3-5",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "mistralai"
  },
  {
   "display_name": "qwen3.8-max",
   "file_slug": "qwen__qwen3.8-max",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "qwen/qwen3.8-max",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "qwen"
  },
  {
   "display_name": "kimi-k3",
   "file_slug": "moonshotai__kimi-k3",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "moonshotai/kimi-k3",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "moonshotai"
  },
  {
   "display_name": "glm-5.2",
   "file_slug": "z-ai__glm-5.2",
   "judge_pair": [
    "openai/gpt-5-mini",
    "google/gemini-3.1-flash-lite"
   ],
   "params": {
    "max_tokens": 1600,
    "max_tokens_regard": 2000,
    "reasoning_effort": "low",
    "repeats": 3,
    "temperature": 0.0,
    "top_p": 1.0
   },
   "slug": "z-ai/glm-5.2",
   "tiebreak": "anthropic/claude-haiku-4.5",
   "vendor": "z-ai"
  }
 ],
 "run_id": "full-2026-08",
 "schema_version": "1",
 "totals": {
  "calls": 75696,
  "cost_usd": 421.088333255,
  "judge_calls": 156068
 }
}
