{
  "source": "https://healthbenchprofessional.com",
  "license": "CC BY 4.0",
  "citation": "Arcophos. HealthBench Professional Leaderboard. 2026-08-16 snapshot. https://healthbenchprofessional.com.",
  "updated": "2026-08-16",
  "benchmark": {
    "name": "HealthBench Professional",
    "scale": "0 to 1",
    "labs": 3,
    "tasks": 525,
    "candidatePool": 15079,
    "physicians": 190,
    "countries": 50,
    "specialties": 26,
    "useCases": 3,
    "grader": "GPT-5.4, low reasoning effort",
    "physicianBaseline": 0.437,
    "paper": "https://arxiv.org/abs/2604.27470",
    "publisher": "OpenAI",
    "released": "2026-04-22",
    "languages": 52,
    "pricingNote": "GPT-5.6 models charge higher rates above 272K input tokens. MAI-Thinking-1 is in public preview on Microsoft Foundry without final list pricing."
  },
  "models": [
    {
      "rank": 1,
      "slug": "claude-fable-5",
      "name": "Claude Fable 5",
      "lab": "Anthropic",
      "logo": "anthropic",
      "score": 0.66,
      "size": null,
      "context": "1.0M",
      "contextTokens": 1000000,
      "inputPer1M": 10,
      "outputPer1M": 50,
      "license": "proprietary",
      "releaseDate": "2026-06-09",
      "positioning": "Anthropic's top-tier model, positioned above the Opus line for the hardest reasoning work."
    },
    {
      "rank": 2,
      "slug": "gpt-5.6-sol",
      "name": "GPT-5.6 Sol",
      "lab": "OpenAI",
      "logo": "openai",
      "score": 0.605,
      "size": null,
      "context": "1.1M",
      "contextTokens": 1100000,
      "inputPer1M": 5,
      "outputPer1M": 30,
      "license": "proprietary",
      "releaseDate": "2026-07-09",
      "positioning": "The flagship tier of OpenAI's GPT-5.6 family, aimed at its hardest reasoning and agentic workloads."
    },
    {
      "rank": 3,
      "slug": "claude-opus-5",
      "name": "Claude Opus 5",
      "lab": "Anthropic",
      "logo": "anthropic",
      "score": 0.598,
      "size": null,
      "context": "1.0M",
      "contextTokens": 1000000,
      "inputPer1M": 5,
      "outputPer1M": 25,
      "license": "proprietary",
      "releaseDate": "2026-07-24",
      "positioning": "Anthropic's frontier workhorse, priced at half of Claude Fable 5."
    },
    {
      "rank": 4,
      "slug": "claude-sonnet-5",
      "name": "Claude Sonnet 5",
      "lab": "Anthropic",
      "logo": "anthropic",
      "score": 0.578,
      "size": null,
      "context": "1.0M",
      "contextTokens": 1000000,
      "inputPer1M": 2,
      "outputPer1M": 10,
      "license": "proprietary",
      "releaseDate": "2026-06-30",
      "positioning": "Anthropic's mid-tier model, positioned for speed at near-Opus quality."
    },
    {
      "rank": 5,
      "slug": "gpt-5.6-terra",
      "name": "GPT-5.6 Terra",
      "lab": "OpenAI",
      "logo": "openai",
      "score": 0.577,
      "size": null,
      "context": "1.1M",
      "contextTokens": 1100000,
      "inputPer1M": 2,
      "outputPer1M": 12,
      "license": "proprietary",
      "releaseDate": "2026-07-09",
      "positioning": "The balanced middle tier of the GPT-5.6 family, between Sol and Luna."
    },
    {
      "rank": 6,
      "slug": "claude-opus-4.8",
      "name": "Claude Opus 4.8",
      "lab": "Anthropic",
      "logo": "anthropic",
      "score": 0.558,
      "size": null,
      "context": "1.0M",
      "contextTokens": 1000000,
      "inputPer1M": 5,
      "outputPer1M": 25,
      "license": "proprietary",
      "releaseDate": "2026-05-28",
      "positioning": "The last release of the Opus 4 series, superseded by Claude Opus 5 in July 2026."
    },
    {
      "rank": 7,
      "slug": "gpt-5.6-luna",
      "name": "GPT-5.6 Luna",
      "lab": "OpenAI",
      "logo": "openai",
      "score": 0.557,
      "size": null,
      "context": "1.1M",
      "contextTokens": 1100000,
      "inputPer1M": 0.2,
      "outputPer1M": 1.2,
      "license": "proprietary",
      "releaseDate": "2026-07-09",
      "positioning": "The small, high-volume tier of the GPT-5.6 family, repriced to $0.20 per million input tokens on July 30, 2026."
    },
    {
      "rank": 8,
      "slug": "gpt-5.5-instant",
      "name": "GPT-5.5 Instant",
      "lab": "OpenAI",
      "logo": "openai",
      "score": 0.384,
      "size": null,
      "context": "400K",
      "contextTokens": 400000,
      "inputPer1M": 5,
      "outputPer1M": 30,
      "license": "proprietary",
      "releaseDate": "2026-05-05",
      "positioning": "The fast variant that became ChatGPT's default model in May 2026."
    },
    {
      "rank": 9,
      "slug": "mai-thinking-1",
      "name": "MAI-Thinking-1",
      "lab": "Microsoft",
      "logo": "microsoft",
      "score": 0.35,
      "size": "1.0T",
      "context": null,
      "contextTokens": null,
      "inputPer1M": null,
      "outputPer1M": null,
      "license": "proprietary",
      "releaseDate": "2026-08-12",
      "positioning": "Microsoft AI's first in-house frontier reasoning model, a sparse mixture-of-experts design in public preview on Microsoft Foundry."
    }
  ],
  "pending": []
}