[
 {
  "candidate": "OpenAI gpt-5.6-sol + web search",
  "id": "openai",
  "runs": 2,
  "score": 97.1,
  "citations/40": 37.1,
  "traps/30": 30.0,
  "coverage/20": 20.0,
  "cost_time/10": 10.0,
  "citation_accuracy": 92.8,
  "claims": 62,
  "supported": 53,
  "partial": 5,
  "unsupported": 2,
  "misattributed": 0,
  "dead": 0,
  "fabricated_src": 0,
  "uncited": 1,
  "unreachable": 0,
  "fabrication_count": 0,
  "traps_per_run": [
   {
    "T1": 10,
    "T2": 10,
    "T3": 10
   },
   {
    "T1": 10,
    "T2": 10,
    "T3": 10
   }
  ],
  "cost_usd_per_run": [
   0.879,
   0.8014
  ],
  "minutes_per_run": [
   2.8,
   2.6
  ],
  "verdict": "HIRED"
 },
 {
  "candidate": "Claude Opus 5.5 + web search/fetch",
  "id": "claude",
  "runs": 2,
  "score": 89.0,
  "citations/40": 35.0,
  "traps/30": 30.0,
  "coverage/20": 17.5,
  "cost_time/10": 6.5,
  "citation_accuracy": 88.7,
  "claims": 114,
  "supported": 94,
  "partial": 9,
  "unsupported": 6,
  "misattributed": 2,
  "dead": 0,
  "fabricated_src": 0,
  "uncited": 1,
  "unreachable": 2,
  "fabrication_count": 4,
  "traps_per_run": [
   {
    "T1": 10,
    "T2": 10,
    "T3": 10
   },
   {
    "T1": 10,
    "T2": 10,
    "T3": 10
   }
  ],
  "cost_usd_per_run": [
   4.9598,
   2.6191
  ],
  "minutes_per_run": [
   6.0,
   2.4
  ],
  "verdict": "HIRED"
 },
 {
  "candidate": "DeepSeek V4 Pro + Search/Fetch (cheap intern)",
  "id": "deepseek",
  "runs": 2,
  "score": 87.4,
  "citations/40": 34.9,
  "traps/30": 27.5,
  "coverage/20": 15.0,
  "cost_time/10": 10.0,
  "citation_accuracy": 87.2,
  "claims": 94,
  "supported": 62,
  "partial": 17,
  "unsupported": 0,
  "misattributed": 2,
  "dead": 0,
  "fabricated_src": 0,
  "uncited": 0,
  "unreachable": 13,
  "fabrication_count": 0,
  "traps_per_run": [
   {
    "T1": 10,
    "T2": 10,
    "T3": 10
   },
   {
    "T1": 5,
    "T2": 10,
    "T3": 10
   }
  ],
  "cost_usd_per_run": [
   0.261,
   0.2731
  ],
  "minutes_per_run": [
   2.6,
   4.8
  ],
  "verdict": "HIRED"
 },
 {
  "candidate": "Gemini Deep Research",
  "id": "gemini",
  "runs": 2,
  "score": 85.6,
  "citations/40": 33.6,
  "traps/30": 27.5,
  "coverage/20": 16.2,
  "cost_time/10": 8.1,
  "citation_accuracy": 86.7,
  "claims": 91,
  "supported": 68,
  "partial": 9,
  "unsupported": 3,
  "misattributed": 2,
  "dead": 0,
  "fabricated_src": 0,
  "uncited": 2,
  "unreachable": 5,
  "fabrication_count": 2,
  "traps_per_run": [
   {
    "T1": 10,
    "T2": 10,
    "T3": 5
   },
   {
    "T1": 10,
    "T2": 10,
    "T3": 10
   }
  ],
  "cost_usd_per_run": [
   2.0447,
   2.9145
  ],
  "minutes_per_run": [
   6.2,
   7.5
  ],
  "verdict": "HIRED"
 }
]