{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "gpt-4-1",
    "name": "GPT-4.1",
    "lab": "openai",
    "date": "2025-04-14",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2025-04-14-gpt-4-1",
    "releases": [],
    "variant_of": null,
    "made_by": "OpenAI",
    "url": "https://frontierlog.ai/model/gpt-4-1",
    "results": [
      {
        "id": "gpt-4-1:gpqa-diamond:default:1",
        "model": "gpt-4-1",
        "model_name": "GPT-4.1",
        "model_date": "2025-04-14",
        "event": "https://frontierlog.ai/event/2025-04-14-gpt-4-1",
        "benchmark": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 66.3,
        "unit": "%",
        "printed": "66.3%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT-4.1",
        "source": "https://openai.com/index/gpt-4-1",
        "source_title": "Introducing GPT-4.1 in the API",
        "source_date": "2025-04-14",
        "where": "academic eval table (columns GPT-4.1 ... GPT-4.5), row 'GPQA Diamond1', column 'GPT-4.1'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "answers parsed by a model, not regex",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.kimi.ai/blog/kimi-k2",
            "name": "Kimi K2",
            "title": "Kimi K2: Open Agentic Intelligence",
            "lab": "moonshot",
            "date": "2025-07-11",
            "printed": "66.3",
            "where": "Kimi-K2-Instruct table, row 'GPQA-Diamond' (metric Avg@8), column 'GPT-4.1'. Unstarred; footnote: '*' marks points taken from the model's tech report/blog, so 66.3 is Moonshot's own run (avg@8, 8k output tokens). Same value on huggingface.co/moonshotai/Kimi-K2-Instruct."
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-for-developers",
            "name": "OpenAI post",
            "title": "Introducing GPT-5 for developers",
            "lab": "openai",
            "date": "2025-08-07",
            "printed": "66.3%",
            "where": "Intelligence table, row 'GPQA diamond (no tools)', column 'GPT-4.1' (answer parsing not stated)"
          },
          {
            "url": "https://www.anthropic.com/news/claude-4",
            "name": "Claude Code generally available",
            "title": "Introducing Claude 4",
            "lab": "anthropic",
            "date": "2025-05-22",
            "printed": "66.3%",
            "where": "HTML table (built from CSV https://cdn.sanity.io/files/4zrzovbb/website/0827c45e02042cf4c6158ea6c04abefedf2b728a.csv, sha256 85f9492bea26e0ac1736ffe36c4007ace95a15df7763754662a71a6a749c07e5), row \"Graduate-level reasoning GPQA Diamond\", column \"OpenAI GPT-4.1\"; sources list cites GPT-4.1 launch post and GPT-4.1 hosted evals"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/b347e2620091fc48.png"
      }
    ]
  }
}