{
  "schemaVersion": "remakebench.subscription-capacity-allowance.v1",
  "evidenceVersion": "v1",
  "id": "claude-max-20x-opus-5-5-2026-09-25",
  "immutable": true,
  "status": "provisional",
  "confidence": {
    "level": "medium",
    "label": "Medium",
    "reason": "Higher confidence: the weekly allowance is measured by dividing corrected in-window consumption by the app-reported 58% quota. It is still one window, one percentage, and mix-dependent."
  },
  "capturedAt": "2026-09-25T09:30:00.000Z",
  "displayDate": "September 25, 2026",
  "provider": {
    "id": "anthropic-claude",
    "displayName": "Anthropic · Claude"
  },
  "plan": {
    "id": "claude-max-20x",
    "displayName": "Claude Max 20×",
    "monthlyPriceUsd": 200,
    "priceSource": "Owner measurement report, 25 September 2026: about 43× the $200 fee.",
    "priceObservedOn": "2026-09-25"
  },
  "model": {
    "id": "claude-opus-5-5",
    "displayName": "Opus 5.5",
    "reasoningMode": "High and Max, as recorded per task",
    "client": "Claude Code"
  },
  "quotaObservations": [
    {
      "id": "claude-max-20x-opus-5-5-2026-09-25-weekly",
      "window": "weekly",
      "observedPercent": 58,
      "observedFraction": 0.58,
      "displayedPrecisionPercentagePoints": 1,
      "resetShown": "2026-10-01 00:59:59 UTC",
      "basis": "base",
      "role": "measurement",
      "source": "Claude app weekly all-models quota, reported 58% used. Window opened 2026-09-24 00:59:59 UTC."
    }
  ],
  "weeklyAllowance": {
    "basis": "measured",
    "basisLabel": "Measured",
    "term": "Weekly all-models allowance, API-equivalent",
    "pointUsd": "2155",
    "pointDisplay": "~$2,155",
    "band": {
      "lowerUsd": "2136",
      "upperUsd": "2174",
      "display": "$2,136 – $2,174",
      "basis": "Rounding band on the displayed 58% (57.5–58.5%). Display precision only, not the full uncertainty."
    },
    "derivation": "Corrected in-window consumption $1,249.83 ÷ 0.58 reported = $2,155 / week. Rounding band on the displayed 58% (57.5–58.5%): $2,136 – $2,174."
  },
  "monthlyAllowance": {
    "basis": "four_week_projection",
    "multiplier": 4,
    "label": "×4 weekly allowance, projected",
    "pointUsd": "8620",
    "pointDisplay": "~$8,620",
    "note": "Monthly (× 4) is a projection of the measured weekly allowance, not a separately metered monthly quota. The report states ~$8,620, about 43× the $200 fee."
  },
  "cohort": {
    "label": "Opus 5.5 grid, 18 tasks, all attempts",
    "runsRecorded": 24,
    "runsRequested": 18,
    "runsDisplay": "24 runs",
    "tasks": 18,
    "apiEquivalentUsd": "1267.39",
    "apiEquivalentUsdDisplay": "$1,267.39",
    "creditEquivalents": null,
    "creditEquivalentsDisplay": null,
    "shareOfWeeklyAllowance": "58.8%",
    "shareBasis": "Harness api_equivalent_cost / api_list_price_equivalent_usd for 24 runs across 18 grid tasks, divided by the $2,155 weekly allowance. The report states $1,267.39 = 58.8% of one week.",
    "tokens": [
      {
        "label": "Courtyard Blossom Stage 2, two attempts",
        "display": "1.49 billion"
      }
    ],
    "tokenNote": "The source report does not publish a grid-wide token total. Courtyard Blossom Stage 2's two attempts moved 1.49 billion tokens between them."
  },
  "method": "Divide corrected in-window API-equivalent consumption ($1,249.83) by the app-reported 58% of the weekly all-models quota. The window opened 2026-09-24 00:59:59 UTC and resets 2026-10-01 00:59:59 UTC. One record per message.id: input and cache counted once, output_tokens taken as the max across a message's rows. Costs are the harness's own api_equivalent_cost / api_list_price_equivalent_usd figures, not recomputed. Monthly allowance is that weekly figure ×4, a projection.",
  "taskCosts": [
    {
      "task": "Courtyard Blossom Stage 2",
      "effort": "High",
      "runs": 2,
      "usdDisplay": "$742.73",
      "shareOfWeek": "34.5%",
      "onGrid": true
    },
    {
      "task": "Oasis Outpost",
      "effort": "High",
      "runs": 2,
      "usdDisplay": "$134.98",
      "shareOfWeek": "6.3%",
      "onGrid": true
    },
    {
      "task": "Sekiro Main Menu",
      "effort": "Max",
      "runs": 2,
      "usdDisplay": "$105.04",
      "shareOfWeek": "4.9%",
      "onGrid": true
    },
    {
      "task": "Jungle Temple",
      "effort": "High",
      "runs": 2,
      "usdDisplay": "$64.99",
      "shareOfWeek": "3.0%",
      "onGrid": true
    },
    {
      "task": "Carve WebGPU Snowboarding",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$57.68",
      "shareOfWeek": "2.7%",
      "onGrid": true
    },
    {
      "task": "Space Flight Stage 2",
      "effort": "High",
      "runs": 1,
      "usdDisplay": "$33.62",
      "shareOfWeek": "1.6%",
      "onGrid": true
    },
    {
      "task": "Neon Gunner",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$33.38",
      "shareOfWeek": "1.5%",
      "onGrid": true
    },
    {
      "task": "Infinite Cathedral",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$28.85",
      "shareOfWeek": "1.3%",
      "onGrid": true
    },
    {
      "task": "Claw Unlock",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$17.39",
      "shareOfWeek": "0.8%",
      "onGrid": true
    },
    {
      "task": "Turbofan (CAD-001)",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$14.07",
      "shareOfWeek": "0.7%",
      "onGrid": true
    },
    {
      "task": "Neo-Gothic Storm City",
      "effort": "High",
      "runs": 3,
      "usdDisplay": "$8.95",
      "shareOfWeek": "0.4%",
      "onGrid": true
    },
    {
      "task": "Bluetooth Speaker PCB",
      "effort": "High",
      "runs": 1,
      "usdDisplay": "$6.47",
      "shareOfWeek": "0.3%",
      "onGrid": true
    },
    {
      "task": "JRPG Boss Battle",
      "effort": "High",
      "runs": 1,
      "usdDisplay": "$6.17",
      "shareOfWeek": "0.3%",
      "onGrid": true
    },
    {
      "task": "Dex Cube Three Moves",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$4.75",
      "shareOfWeek": "0.2%",
      "onGrid": true
    },
    {
      "task": "Dex Cube Turn",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$2.20",
      "shareOfWeek": "0.1%",
      "onGrid": true
    },
    {
      "task": "Apple Stem",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$2.11",
      "shareOfWeek": "0.1%",
      "onGrid": true
    },
    {
      "task": "Eiffel Drawing",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$2.01",
      "shareOfWeek": "0.1%",
      "onGrid": true
    },
    {
      "task": "Apple Carry",
      "effort": "Max",
      "runs": 1,
      "usdDisplay": "$2.00",
      "shareOfWeek": "0.1%",
      "onGrid": true
    },
    {
      "task": "Total (18 grid tasks)",
      "effort": "",
      "runs": 24,
      "usdDisplay": "$1,267.39",
      "shareOfWeek": "58.8%",
      "onGrid": true
    },
    {
      "task": "Space Flight Stage 1 (off-grid)",
      "effort": "High",
      "runs": 1,
      "usdDisplay": "$13.88",
      "shareOfWeek": "0.6%",
      "onGrid": false
    },
    {
      "task": "All Opus 5.5 work",
      "effort": "",
      "runs": 25,
      "usdDisplay": "$1,281.27",
      "shareOfWeek": "59.5%",
      "onGrid": false
    }
  ],
  "windowConsumption": {
    "reportedPercentUsed": "58%",
    "resetAt": "2026-10-01 00:59:59 UTC",
    "windowOpenedAt": "2026-09-24 00:59:59 UTC",
    "components": [
      {
        "label": "Local Claude Code transcripts, all models, in-window",
        "usd": "$648.51"
      },
      {
        "label": "Oasis Outpost a1 (repo receipt, absent from local)",
        "usd": "$91.43"
      },
      {
        "label": "Jungle Temple a1",
        "usd": "$30.01"
      },
      {
        "label": "Sekiro high a2",
        "usd": "$11.17"
      },
      {
        "label": "Courtyard Blossom a1",
        "usd": "$433.73"
      },
      {
        "label": "Jungle Temple a2",
        "usd": "$34.98"
      }
    ],
    "correctedUsd": "$1,249.83",
    "formula": "$1,249.83 ÷ 0.58 reported = $2,155 / week"
  },
  "rateCard": {
    "model": "Opus 5.5",
    "usdPerMillion": {
      "input": "4.00",
      "cacheWrite5m": "5.00",
      "cacheWrite1h": "8.00",
      "cacheRead": "0.20",
      "output": "20.00"
    },
    "note": "Solved by least squares from seven independent harness receipts, then confirmed against the card the repo records with verified_live_pricing: true. Write multipliers land on 1.25× and 2.00×. The card reproduces all seven receipts to 0.0000% error. Opus 5.5 is ~20% cheaper per token than Opus 5 ($5/$25, cache read $0.50)."
  },
  "supportingObservations": [
    {
      "label": "Remaining this week (42%)",
      "value": "~$905"
    },
    {
      "label": "Fee multiple",
      "value": "about 43× the $200 fee"
    },
    {
      "label": "Opus 5.5 benchmark work inside this window",
      "value": "$1,203.37, 96.3% of the $1,249.83 consumed, 55.8 of the reported 58 points"
    },
    {
      "label": "Prior-window Opus 5.5 benchmark work",
      "value": "$77.90 (the 09-22/23 robotics and CAD runs)"
    },
    {
      "label": "Grid at stated effort, dropping the Sekiro high retry",
      "value": "$1,256.22, or 58.3%"
    },
    {
      "label": "Courtyard Blossom Stage 2 share of the grid",
      "value": "59% of the grid's cost; the failed first attempt alone was $433.73"
    },
    {
      "label": "Eight robotics tasks",
      "value": "$30.46, 2.4% of the grid"
    },
    {
      "label": "Local transcripts versus harness receipts",
      "value": "Local transcripts capture $756 of the $1,281 of Opus 5.5 work, 59%. Local data alone would have put the allowance at $1,118."
    },
    {
      "label": "Opus-5-dominated week of Aug 20–27",
      "value": "$4,101"
    },
    {
      "label": "Extra usage already spent",
      "value": "$72.41"
    }
  ],
  "caveats": [
    "Mix dependence is the dominant term, and it is large. The quota meters weighted units, not dollars, so this $2,155 is denominated in an Opus-5.5-heavy week (96% of the window). Aligned to the same reset boundary, the Opus-5-dominated week of Aug 20–27 consumed $4,101 — nearly double this allowance, in a week that may not even have hit 100%. Opus 5.5 therefore burns roughly 2× more quota per API dollar than Opus 5. Quote $2,155 for Opus-5.5 work; expect closer to $4,000+ for Opus-5 work.",
    "The plan string reads \"Pro\", not Max 20x. The app reports plan: \"Pro\". If that is literal rather than a coarse family label, this is a Pro allowance and the Max 20x figure would be higher. The same ambiguity appeared in the Codex audit. Worth confirming in billing.",
    "Extra usage: $72.41 already spent. Overage is currently disabled but has been charged before, so some past weeks ran past their cap — meaning historical weekly totals are not clean ceiling observations.",
    "Capture may still be incomplete. Five missing runs were recoverable because the harness had receipts. Non-benchmark Claude work outside local transcripts (claude.ai web, mobile, cloud) would be invisible and has no receipt. That pushes the allowance up, not down. Local transcripts capture only $756 of the $1,281 of Opus 5.5 work the harness has receipts for (59%). Using local data alone would have put the allowance at $1,118.",
    "One window, one percentage. This is a single observation at one point in one window. It is far better evidence than the September 6 inference, but it is not a distribution."
  ],
  "evidence": {
    "basePath": "/evidence/subscription-capacity-index/claude-max-20x/v1",
    "measurementUrl": "/evidence/subscription-capacity-index/claude-max-20x/v1/measurement.json",
    "methodologyUrl": "/evidence/subscription-capacity-index/claude-max-20x/v1/report.md",
    "manifestUrl": "/evidence/subscription-capacity-index/claude-max-20x/v1/manifest.json",
    "checksumsUrl": "/evidence/subscription-capacity-index/claude-max-20x/v1/checksums.sha256",
    "resultsUrl": "https://www.remakebench.com/results",
    "reportUrl": "/evidence/subscription-capacity-index/claude-max-20x/v1/report.md",
    "scriptUrls": [
      "/evidence/subscription-capacity-index/claude-max-20x/v1/audit.py",
      "/evidence/subscription-capacity-index/claude-max-20x/v1/allow.py",
      "/evidence/subscription-capacity-index/claude-max-20x/v1/hist.py",
      "/evidence/subscription-capacity-index/claude-max-20x/v1/sess.py"
    ]
  },
  "presentation": {
    "versionLabel": "Subscription capacity index · 25 September 2026",
    "title": "Claude Max 20×. Weekly allowance ~$2,155 API-equivalent.",
    "deck": "Measured from the app's reported 58% of the weekly all-models quota. Monthly is that weekly figure ×4, a projection. The plan string reads Pro.",
    "seoDescription": "Claude Max 20× weekly allowance ~$2,155 API-equivalent ($2,136–$2,174), monthly projection ~$8,620, with the Opus 5.5 grid at 58.8% of one week.",
    "planStringCaveat": "The plan string reads \"Pro\", not Max 20×. The app reports plan: \"Pro\". If that is literal rather than a coarse family label, this is a Pro allowance and the Max 20× figure would be higher."
  }
}