{
  "scope": "Solver inference only, API-equivalent at list rates; the model ran on a ChatGPT Pro subscription through Codex, so these are estimates, not bills. Judging and local infrastructure are excluded. 109 of 115 runs are priced; the six runs that hit the 3-hour bound ended before Codex reported usage and are unpriced, not $0, so extra-high and max spend is understated. Runtime covers all 115 runs.",
  "pricing_verified": "2026-09-25",
  "sources": {
    "openai": "https://developers.openai.com/api/docs/pricing"
  },
  "rates_per_million": {
    "gpt6luna": {
      "input": 0.1,
      "cached_input": 0.01,
      "output": 0.5
    }
  },
  "groups": [
    {
      "model": "gpt6luna",
      "effort": "low",
      "n": 23,
      "runs": 23,
      "unpriced_timeouts": 0,
      "usd": {
        "n": 23,
        "mean": 0.027790478260869566,
        "se": 0.007274311815342826
      },
      "usd_total": 0.639181,
      "raw_tokens": {
        "n": 23,
        "mean": 1654399.3043478262,
        "se": 551511.7654505488
      },
      "raw_tokens_total": 38051184,
      "minutes": {
        "n": 23,
        "mean": 5.557563043478261,
        "se": 1.047903222073788
      },
      "solver_fallback_runs": 0
    },
    {
      "model": "gpt6luna",
      "effort": "medium",
      "n": 23,
      "runs": 23,
      "unpriced_timeouts": 0,
      "usd": {
        "n": 23,
        "mean": 0.009534565217391305,
        "se": 0.0006128070346880734
      },
      "usd_total": 0.21929500000000002,
      "raw_tokens": {
        "n": 23,
        "mean": 404296.52173913043,
        "se": 23814.097299897192
      },
      "raw_tokens_total": 9298820,
      "minutes": {
        "n": 23,
        "mean": 3.559326811594203,
        "se": 0.3131111633485139
      },
      "solver_fallback_runs": 0
    },
    {
      "model": "gpt6luna",
      "effort": "high",
      "n": 23,
      "runs": 23,
      "unpriced_timeouts": 0,
      "usd": {
        "n": 23,
        "mean": 0.05191208695652174,
        "se": 0.005020368430456266
      },
      "usd_total": 1.193978,
      "raw_tokens": {
        "n": 23,
        "mean": 2556496.695652174,
        "se": 303930.9423606601
      },
      "raw_tokens_total": 58799424,
      "minutes": {
        "n": 23,
        "mean": 15.590407246376811,
        "se": 1.67257649785497
      },
      "solver_fallback_runs": 0
    },
    {
      "model": "gpt6luna",
      "effort": "extra-high",
      "n": 21,
      "runs": 23,
      "unpriced_timeouts": 2,
      "usd": {
        "n": 21,
        "mean": 0.15968628571428573,
        "se": 0.03641089140138617
      },
      "usd_total": 3.353412,
      "raw_tokens": {
        "n": 21,
        "mean": 8473100.904761905,
        "se": 2204760.8680581003
      },
      "raw_tokens_total": 177935119,
      "minutes": {
        "n": 23,
        "mean": 52.57451956521739,
        "se": 10.579109521244105
      },
      "solver_fallback_runs": 0
    },
    {
      "model": "gpt6luna",
      "effort": "max",
      "n": 19,
      "runs": 23,
      "unpriced_timeouts": 4,
      "usd": {
        "n": 19,
        "mean": 0.09752915789473685,
        "se": 0.01412181073227736
      },
      "usd_total": 1.853054,
      "raw_tokens": {
        "n": 19,
        "mean": 4317174.052631579,
        "se": 758489.5504104565
      },
      "raw_tokens_total": 82026307,
      "minutes": {
        "n": 23,
        "mean": 60.53421449275363,
        "se": 12.55429656356173
      },
      "solver_fallback_runs": 0
    }
  ],
  "totals": {
    "gpt6luna": {
      "runs": 115,
      "priced_runs": 109,
      "unpriced_timeouts": 6,
      "usd": 7.25892,
      "raw_tokens": 366110854,
      "solver_hours": 52.82947861111111
    }
  },
  "unpriced": [
    {
      "effort": "extra-high",
      "task": "legacy-depotcore-binary-parity",
      "duration_s": 10800.631
    },
    {
      "effort": "extra-high",
      "task": "legacy-paddockcore-binary-parity",
      "duration_s": 10800.65
    },
    {
      "effort": "max",
      "task": "legacy-cellarcore-binary-parity",
      "duration_s": 10800.638
    },
    {
      "effort": "max",
      "task": "legacy-depotcore-binary-parity",
      "duration_s": 10800.656
    },
    {
      "effort": "max",
      "task": "legacy-lodgecore-binary-parity",
      "duration_s": 10800.664
    },
    {
      "effort": "max",
      "task": "legacy-paddockcore-binary-parity",
      "duration_s": 10800.611
    }
  ],
  "limitations": [
    "Codex receipts report input, cached input, output and reasoning tokens per run; cached input is billed at the cache-read rate and the rest of the input at the standard rate. Cache writes are free on this API and are not modelled.",
    "Standard tier, short-context rates. OpenAI bills prompts over 272K input tokens at a higher rate, but Codex keeps each request within its 272K context window, so no long-context premium applies and none is modelled.",
    "No Batch, Flex, Fast or priority pricing is applied.",
    "The six timed-out runs have no usage receipt. They are unpriced and excluded from cost and token means and totals; level means cover priced runs only, so extra-high and max spend is understated.",
    "The estimate covers the solver's exposed receipts, not an independently observed API invoice.",
    "The sweep stamped each run at run time at the published GPT-6 Luna list rates; the export recomputes every priced run from its receipt and matched every stamp."
  ],
  "comparison_sha256": "6cfc0dc8f703992a247d95d2132f125862bdc4807e5c85344e779c3b248b3a31",
  "note": "Per-run estimates are in runs.json (estimated_usd, raw_tokens, token_usage). Judging is excluded."
}