{
  "test": "quality-done",
  "title": "Quality of finished work: Right when done",
  "definition": "Share of eligible runs in which every required item checked out, among runs where the agent said it was done.",
  "denominator": "Runs that said done. Pooled counts: 317 without USDC; 573 with USDC. Runs that gave up or exhausted their budget are excluded.",
  "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": "Scores are fractions from 0 to 1. Changes ending in _pp are percentage points; counts are eligible runs or claims as defined above.",
  "source": {
    "study": "/experiments/funded-agent-index",
    "methodology": "/experiments/funded-agent-index#spend-top"
  },
  "chartUrl": "/experiments/funded-agent-index?quality=done#quality",
  "downloads": {
    "json": "/experiments/funded-agent-index/data/quality-done.json",
    "csv": "/experiments/funded-agent-index/data/quality-done.csv"
  },
  "rows": [
    {
      "model": "Claude Opus 5.5",
      "model_id": "claude-opus-5-5",
      "without_usdc": 0.75,
      "with_usdc": 0.8526315789473684,
      "change_pp": 10.26315789473684,
      "without_eligible_count": 52,
      "with_eligible_count": 95
    },
    {
      "model": "Claude Sonnet 5.5",
      "model_id": "claude-sonnet-5-5",
      "without_usdc": 0.7708333333333334,
      "with_usdc": 0.8444444444444444,
      "change_pp": 7.361111111111107,
      "without_eligible_count": 48,
      "with_eligible_count": 90
    },
    {
      "model": "Claude Haiku 5.5",
      "model_id": "claude-haiku-5-5",
      "without_usdc": 0.6666666666666666,
      "with_usdc": 0.7288135593220338,
      "change_pp": 6.214689265536721,
      "without_eligible_count": 30,
      "with_eligible_count": 59
    },
    {
      "model": "Grok 4.6",
      "model_id": "grok-4.6",
      "without_usdc": 0.54,
      "with_usdc": 0.7790697674418605,
      "change_pp": 23.906976744186046,
      "without_eligible_count": 50,
      "with_eligible_count": 86
    },
    {
      "model": "GPT-6 Luna",
      "model_id": "gpt-6-luna",
      "without_usdc": 0.5945945945945946,
      "with_usdc": 0.7397260273972602,
      "change_pp": 14.513143280266561,
      "without_eligible_count": 37,
      "with_eligible_count": 73
    },
    {
      "model": "GPT-6.1 Sol",
      "model_id": "gpt-6.1-sol",
      "without_usdc": 0.8478260869565217,
      "with_usdc": 0.8444444444444444,
      "change_pp": -0.33816425120772875,
      "without_eligible_count": 46,
      "with_eligible_count": 90
    },
    {
      "model": "Gemini 3 Flash",
      "model_id": "gemini-3-flash-preview",
      "without_usdc": 0.5925925925925926,
      "with_usdc": 0.6875,
      "change_pp": 9.490740740740744,
      "without_eligible_count": 54,
      "with_eligible_count": 80
    },
    {
      "model": "All models (pooled)",
      "model_id": "all",
      "without_usdc": 0.6813880126182965,
      "with_usdc": 0.7888307155322862,
      "change_pp": 10.744270291398973,
      "without_eligible_count": 317,
      "with_eligible_count": 573
    }
  ],
  "taskRows": []
}