{
  "test": "study-spend",
  "title": "Study spend by model",
  "definition": "Research expense for both conditions combined: model inference, agent purchases, checker purchases and web search. This ledger is distinct from the model-plus-service comparison.",
  "denominator": "Seven models, 21 tasks, five repeats per condition; 1,470 runs in total. Completion and cost give tasks and models equal weight.",
  "snapshot": "2026-11-13 16:59 UTC",
  "scope": "Seven models, 21 tasks, five repeats per condition; 1,470 runs in total. Completion and cost give tasks and models equal weight.",
  "availability": "These are the published run aggregates, not individual run logs. The source snapshot does not include per-run records, prompts, outputs or traces. No individual run records have been reconstructed.",
  "units": "Costs are USD. Completion scores are fractions from 0 to 1; changes ending in _pp are percentage points.",
  "source": {
    "study": "/experiments/funded-agent-index",
    "methodology": "/experiments/funded-agent-index#spend-top"
  },
  "chartUrl": "/experiments/funded-agent-index#spend",
  "downloads": {
    "json": "/experiments/funded-agent-index/data/study-spend.json",
    "csv": "/experiments/funded-agent-index/data/study-spend.csv"
  },
  "rows": [
    {
      "model": "Claude Opus 5.5",
      "model_id": "claude-opus-5-5",
      "model_usd": 34.6931,
      "agent_purchases_usd": 6.846,
      "checker_purchases_usd": 0.82,
      "web_search_usd": 1.764,
      "total_usd": 44.12310000000001,
      "runs": 210
    },
    {
      "model": "Claude Sonnet 5.5",
      "model_id": "claude-sonnet-5-5",
      "model_usd": 18.9256,
      "agent_purchases_usd": 6.0145,
      "checker_purchases_usd": 0.94,
      "web_search_usd": 2.947,
      "total_usd": 28.8271,
      "runs": 210
    },
    {
      "model": "Claude Haiku 5.5",
      "model_id": "claude-haiku-5-5",
      "model_usd": 1.0304,
      "agent_purchases_usd": 3.3895,
      "checker_purchases_usd": 0.8,
      "web_search_usd": 3.185,
      "total_usd": 8.4049,
      "runs": 210
    },
    {
      "model": "Grok 4.6",
      "model_id": "grok-4.6",
      "model_usd": 41.8402,
      "agent_purchases_usd": 6.7957,
      "checker_purchases_usd": 0.82,
      "web_search_usd": 10.241,
      "total_usd": 59.69690000000001,
      "runs": 210
    },
    {
      "model": "GPT-6 Luna",
      "model_id": "gpt-6-luna",
      "model_usd": 1.4159,
      "agent_purchases_usd": 6.4686,
      "checker_purchases_usd": 0.84,
      "web_search_usd": 13.888,
      "total_usd": 22.6125,
      "runs": 210
    },
    {
      "model": "GPT-6.1 Sol",
      "model_id": "gpt-6.1-sol",
      "model_usd": 11.8682,
      "agent_purchases_usd": 8.9744,
      "checker_purchases_usd": 1.12,
      "web_search_usd": 3.227,
      "total_usd": 25.1896,
      "runs": 210
    },
    {
      "model": "Gemini 3 Flash",
      "model_id": "gemini-3-flash-preview",
      "model_usd": 26.4847,
      "agent_purchases_usd": 7.8511,
      "checker_purchases_usd": 1.5,
      "web_search_usd": 13.587,
      "total_usd": 49.422799999999995,
      "runs": 210
    },
    {
      "model": "All models",
      "model_id": "all",
      "model_usd": 136.26,
      "agent_purchases_usd": 46.34,
      "checker_purchases_usd": 6.84,
      "web_search_usd": 48.84,
      "total_usd": 238.28,
      "runs": 1470
    }
  ],
  "taskRows": []
}