{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "gpt-5-4",
    "name": "GPT-5.4",
    "lab": "openai",
    "date": "2026-03-05",
    "flagship": true,
    "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
    "releases": [],
    "variant_of": null,
    "made_by": "OpenAI",
    "url": "https://frontierlog.ai/model/gpt-5-4",
    "results": [
      {
        "id": "gpt-5-4:terminal-bench-2-0:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "terminal-bench-2-0",
        "benchmark_name": "Terminal-Bench 2.0",
        "setting": "default",
        "setting_name": "",
        "score": 75.1,
        "unit": "%",
        "printed": "75.1%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "coding table, row 'Terminal-Bench 2.0', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "harness not stated",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "75.1%",
            "where": "Table 8.1.A (p.191), row Terminal-Bench 2.0, column GPT-5.4 (footnote 38: OpenAI's specialized harness)"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "75.1%",
            "where": "benchmark table image, row Terminal-Bench 2.0, column GPT-5.4 (labelled 'self-reported harness')"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "75.1",
            "where": "Bar chart image; group 'Terminal Bench 2.0 (Acc)'; value label on GPT-5.4-xHigh bar"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "75.1%",
            "where": "table 0 \"Terminal-Bench 2.0\" row, \"GPT-5.4\" column"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
            "name": "GPT-5.4 mini and nano",
            "title": "Introducing GPT-5.4 mini and nano",
            "lab": "openai",
            "date": "2026-03-17",
            "printed": "75.1%",
            "where": "table 0 \"Terminal-Bench 2.0\" row, \"GPT-5.4 (xhigh)\" column"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "75.1",
            "where": "Benchmark table, row 'Terminal-Bench 2.0 Best self-reported harness', column 'GPT-5.4' (Codex)"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/9941c2d6c8891638.png"
      },
      {
        "id": "gpt-5-4:gpqa-diamond:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 92.8,
        "unit": "%",
        "printed": "92.8%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "academic table, row 'GPQA Diamond', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "xhigh effort",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "92.8%",
            "where": "Table 8.1.A (p.191), row GPQA Diamond, column GPT-5.4"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "92.8",
            "where": "Benchmark Table (appendix), row GPQA-Diamond ; column GPT-5.4 (xhigh)"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "92.8%",
            "where": "table 5 \"GPQA Diamond\" row, \"GPT-5.4\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/2322d7fe4c245f75.png"
      },
      {
        "id": "gpt-5-4:hle:no-tools:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 39.8,
        "unit": "%",
        "printed": "39.8%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "academic table, row 'Humanity's Last Exam (no tools)', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "39.8%",
            "where": "Table 8.1.A (p.191), row Humanity's Last Exam / No tools, column GPT-5.4"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "39.8",
            "where": "Bar chart image; group 'HLE (Pass@1), under 'Knowledge & Reasoning''; value label on GPT-5.4-xHigh bar"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "39.8",
            "where": "Benchmark Table (appendix), row HLE-Full (Reasoning & Knowledge; no tools) ; column GPT-5.4 (xhigh)"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "39.8%",
            "where": "table 5 \"Humanity's Last Exam (no tools)\" row, \"GPT-5.4\" column"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
            "name": "GPT-5.4 mini and nano",
            "title": "Introducing GPT-5.4 mini and nano",
            "lab": "openai",
            "date": "2026-03-17",
            "printed": "39.8%",
            "where": "table 3 \"HLE w/o tools\" row, \"GPT-5.4 (xhigh)\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/2322d7fe4c245f75.png"
      },
      {
        "id": "gpt-5-4:arc-agi-2:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "arc-agi-2",
        "benchmark_name": "ARC-AGI-2",
        "setting": "default",
        "setting_name": "",
        "score": 73.3,
        "unit": "%",
        "printed": "73.3%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "abstract reasoning table, row 'ARC-AGI-2 (Verified)', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "ARC Prize Verified",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-16",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "73.3%",
            "where": "Table 8.1.A (p.192), row ARC-AGI-2, column GPT-5.4"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "73.3%",
            "where": "table 8 \"ARC-AGI-2 (Verified)\" row, \"GPT-5.4\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/164992c9e384e828.png"
      },
      {
        "id": "gpt-5-4:hle:with-tools:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 52.1,
        "unit": "%",
        "printed": "52.1%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "academic eval table, column 'GPT‑5.4', row 'Humanity's Last Exam (with tools)'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "52.1%",
            "where": "Table 8.1.A (p.191), row Humanity's Last Exam / With tools, column GPT-5.4"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "52.1",
            "where": "Benchmark Table (appendix), row HLE-Full w/ tools ; column GPT-5.4 (xhigh)"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "52.1%",
            "where": "table 5 \"Humanity's Last Exam (with tools)\" row, \"GPT-5.4\" column"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
            "name": "GPT-5.4 mini and nano",
            "title": "Introducing GPT-5.4 mini and nano",
            "lab": "openai",
            "date": "2026-03-17",
            "printed": "52.1%",
            "where": "table 3 \"HLE w/ tool\" row, \"GPT-5.4 (xhigh)\" column"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "52.1",
            "where": "Benchmark table, row 'HLE w/ Tools (* = full set for Opus 4.6, Gemini 3.1 Pro, GPT-5.4)', column 'GPT-5.4'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/2322d7fe4c245f75.png"
      },
      {
        "id": "gpt-5-4:browsecomp:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "browsecomp",
        "benchmark_name": "BrowseComp",
        "setting": "default",
        "setting_name": "",
        "score": 82.7,
        "unit": "%",
        "printed": "82.7%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "tool-use eval table (columns GPT‑5.4 / GPT‑5.4Pro / GPT‑5.3-Codex / GPT‑5.2 / GPT‑5.2Pro), row 'BrowseComp', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "82.7%",
            "where": "Table 8.1.A (p.191), row BrowseComp, column GPT-5.4"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "82.7",
            "where": "Full benchmark table image; row 'BrowseComp (Pass@1)'; column 'GPT-5.4 xHigh'"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "82.7",
            "where": "Benchmark Table (appendix), row BrowseComp ; column GPT-5.4 (xhigh)"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "82.7%",
            "where": "table 0 \"BrowseComp\" row, \"GPT-5.4\" column"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "82.7",
            "where": "Benchmark table, row 'BrowseComp w/ Context Manage', column 'GPT-5.4'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/447282185a6a0973.png"
      },
      {
        "id": "gpt-5-4:swe-bench-pro:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "swe-bench-pro",
        "benchmark_name": "SWE-bench Pro",
        "setting": "default",
        "setting_name": "",
        "score": 57.7,
        "unit": "%",
        "printed": "57.7%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "coding table (columns GPT‑5.4, GPT‑5.4 Pro, GPT‑5.3-Codex, …), row 'SWE-Bench Pro (Public)', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "57.7%",
            "where": "Table 8.1.A (p.191), row SWE-bench Pro, column GPT-5.4"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "57.7%",
            "where": "benchmark table image, row SWE-bench Pro, column GPT-5.4"
          },
          {
            "url": "https://api-docs.deepseek.com/news/news260424/",
            "name": "DeepSeek V4 Preview",
            "title": "DeepSeek V4 Preview Release",
            "lab": "deepseek",
            "date": "2026-04-24",
            "printed": "57.7",
            "where": "Full benchmark table image; row 'SWE Pro (Resolved)'; column 'GPT-5.4 xHigh'"
          },
          {
            "url": "https://www.minimax.io/news/minimax-m27-en",
            "name": "MiniMax M2.7",
            "title": "MiniMax M2.7: Early Echoes of Self-Evolution",
            "lab": "minimax",
            "date": "2026-03-18",
            "printed": "57.7",
            "where": "Bar chart image (8 panels); panel 'SWE Bench Pro'; value label on GPT 5.4 bar"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "57.7",
            "where": "Benchmark Table (appendix), row SWE-Bench Pro ; column GPT-5.4 (xhigh)"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "57.7%",
            "where": "table 1 \"SWE-Bench Pro (Public) *\" row, \"GPT-5.4\" column"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
            "name": "GPT-5.4 mini and nano",
            "title": "Introducing GPT-5.4 mini and nano",
            "lab": "openai",
            "date": "2026-03-17",
            "printed": "57.7%",
            "where": "table 0 \"SWE-Bench Pro (Public)\" row, \"GPT-5.4 (xhigh)\" column"
          },
          {
            "url": "https://z.ai/blog/glm-5.1",
            "name": "GLM-5.1",
            "title": "GLM-5.1: Towards Long-Horizon Tasks",
            "lab": "zai",
            "date": "2026-04-07",
            "printed": "57.7",
            "where": "Tabbed bar chart at top, tab 'Software Engineering' (SWE-Bench Pro), bar 'GPT-5.4'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/9941c2d6c8891638.png"
      },
      {
        "id": "gpt-5-4:osworld-verified:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "osworld-verified",
        "benchmark_name": "OSWorld-Verified",
        "setting": "default",
        "setting_name": "",
        "score": 75.0,
        "unit": "%",
        "printed": "75.0%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "headline eval table (HTML, first table), row 'OSWorld-Verified', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-18",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "75.0%",
            "where": "Table 8.1.A (p.191), row OSWorld, column GPT-5.4 (OSWorld-Verified per Figure 8.9.4.A)"
          },
          {
            "url": "https://www.anthropic.com/news/claude-opus-4-7",
            "name": "Claude Opus 4.7",
            "title": "Introducing Claude Opus 4.7",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "75.0%",
            "where": "benchmark table image, row OSWorld-Verified, column GPT-5.4"
          },
          {
            "url": "https://www.kimi.ai/blog/kimi-k2-6",
            "name": "Kimi K2.6",
            "title": "Kimi K2.6: Advancing Open-Source Coding",
            "lab": "moonshot",
            "date": "2026-04-20",
            "printed": "75.0",
            "where": "Benchmark Table (appendix), row OSWorld-Verified ; column GPT-5.4 (xhigh)"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "75.0%",
            "where": "table 0 \"OSWorld-Verified\" row, \"GPT-5.4\" column"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
            "name": "GPT-5.4 mini and nano",
            "title": "Introducing GPT-5.4 mini and nano",
            "lab": "openai",
            "date": "2026-03-17",
            "printed": "75.0%",
            "where": "table 0 \"OSWorld-Verified\" row, \"GPT-5.4 (xhigh)\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/447282185a6a0973.png"
      },
      {
        "id": "gpt-5-4:arc-agi-1:default:1",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "arc-agi-1",
        "benchmark_name": "ARC-AGI-1",
        "setting": "default",
        "setting_name": "",
        "score": 93.7,
        "unit": "%",
        "printed": "93.7%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT‑5.4",
        "source": "https://openai.com/index/introducing-gpt-5-4",
        "source_title": "Introducing GPT-5.4",
        "source_date": "2026-03-05",
        "where": "detail eval table 'ARC-AGI' (HTML), row 'ARC-AGI-1 (Verified)', column 'GPT‑5.4'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "ARC Prize Verified",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-18",
        "check": "human",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-opus-4-7-system-card",
            "name": "Claude Opus 4.7 card",
            "title": "Claude Opus 4.7 System Card",
            "lab": "anthropic",
            "date": "2026-04-16",
            "printed": "93.7%",
            "where": "Table 8.1.A (p.192), row ARC-AGI-1, column GPT-5.4"
          },
          {
            "url": "https://openai.com/index/introducing-gpt-5-5",
            "name": "GPT-5.5",
            "title": "Introducing GPT-5.5",
            "lab": "openai",
            "date": "2026-04-23",
            "printed": "93.7%",
            "where": "table 8 \"ARC-AGI-1 (Verified)\" row, \"GPT-5.4\" column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/164992c9e384e828.png"
      },
      {
        "id": "gpt-5-4:gpqa-diamond:default:2",
        "model": "gpt-5-4",
        "model_name": "GPT-5.4",
        "model_date": "2026-03-05",
        "event": "https://frontierlog.ai/event/2026-03-05-gpt-5-4",
        "benchmark": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 93.0,
        "unit": "%",
        "printed": "93.0%",
        "reported_by": "openai",
        "run_by": null,
        "label": "GPT-5.4 (xhigh)",
        "source": "https://openai.com/index/introducing-gpt-5-4-mini-and-nano",
        "source_title": "Introducing GPT-5.4 mini and nano",
        "source_date": "2026-03-17",
        "where": "academic table (columns GPT-5.4 (xhigh), GPT-5.4 mini (xhigh), GPT-5.4 nano (xhigh), GPT-5 mini), row 'GPQA Diamond', column 'GPT-5.4 (xhigh)'",
        "own": false,
        "primary": false,
        "parallel": false,
        "notes": "xhigh effort",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-20",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_capture": "https://frontierlog.ai/captures/ec1c7aef5bba5db5.png"
      }
    ]
  }
}