{
  "api_version": "1",
  "updated": "2026-11-10",
  "latest_event": "2026-11-09",
  "rights": "© Frontier Log, all rights reserved.",
  "benchmark": {
    "slug": "hle",
    "name": "Humanity's Last Exam",
    "family": "hle",
    "unit": "%",
    "higher_is_better": true,
    "status": "active",
    "active": true,
    "superseded_by": null,
    "retired": null,
    "benchmark_url": "https://lastexam.ai",
    "introduced": "2025-01-23",
    "creator": "Center for AI Safety and Scale AI",
    "introduced_source": "https://scale.com/blog/humanitys-last-exam-results",
    "settings": [
      {
        "setting": "no-tools",
        "name": "No tools, full set",
        "url": "https://frontierlog.ai/benchmark/hle",
        "results": 38
      },
      {
        "setting": "with-tools",
        "name": "With tools, full set",
        "url": "https://frontierlog.ai/benchmark/hle/with-tools",
        "results": 41
      },
      {
        "setting": "no-tools-text-only",
        "name": "No tools, text-only subset",
        "url": "https://frontierlog.ai/benchmark/hle/no-tools-text-only",
        "results": 29
      },
      {
        "setting": "with-tools-text-only",
        "name": "With tools, text-only subset",
        "url": "https://frontierlog.ai/benchmark/hle/with-tools-text-only",
        "results": 15
      },
      {
        "setting": "no-tools-unstated",
        "name": "No tools, set not stated",
        "url": "https://frontierlog.ai/benchmark/hle/no-tools-unstated",
        "results": 41
      },
      {
        "setting": "with-tools-unstated",
        "name": "With tools, set not stated",
        "url": "https://frontierlog.ai/benchmark/hle/with-tools-unstated",
        "results": 28
      }
    ]
  },
  "total": 41,
  "results": [
    {
      "id": "openai-o3-mini:hle:no-tools-unstated:1",
      "model": "openai-o3-mini",
      "model_name": "o3-mini",
      "model_date": "2025-01-31",
      "event": "https://frontierlog.ai/event/2025-01-31-openai-o3-mini",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 13.4,
      "unit": "%",
      "printed": "13.40",
      "reported_by": "openai",
      "run_by": null,
      "label": "o3-mini",
      "source": "https://openai.com/index/introducing-o3-and-o4-mini",
      "source_title": "Introducing OpenAI o3 and o4-mini",
      "source_date": "2025-04-16",
      "where": "'What's changed' section, bar chart 'Humanity's Last Exam', 2nd bar 'o3-mini' (page screenshot (job 20879 capture), part 02)",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://openai.com/index/introducing-gpt-oss",
          "name": "gpt-oss-120b and gpt-oss-20b",
          "title": "Introducing gpt-oss",
          "lab": "openai",
          "date": "2025-08-05",
          "printed": "13.4",
          "where": "Humanity's Last Exam chart, bar 'o3-mini (without tools)' (effort not stated)"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/392f30ae76e5713f.png"
    },
    {
      "id": "gemini-2-5-pro:hle:no-tools-unstated:1",
      "model": "gemini-2-5-pro",
      "model_name": "Gemini 2.5 Pro",
      "model_date": "2025-03-25",
      "event": "https://frontierlog.ai/event/2025-03-25-gemini-2-5-pro",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 18.8,
      "unit": "%",
      "printed": "18.8%",
      "reported_by": "google",
      "run_by": null,
      "label": "Gemini 2.5 Pro",
      "source": "https://blog.google/innovation-and-ai/models-and-research/google-deepmind/gemini-model-thinking-updates-march-2025/",
      "source_title": "Gemini 2.5: Our most intelligent AI model",
      "source_date": "2025-03-25",
      "where": "benchmark table (image), row 'Reasoning & knowledge, Humanity's Last Exam (no tools)', column 'Gemini 2.5 Pro Experimental (03-25)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "pass@1",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/a8f4323681d1a15d.gif",
      "source_image_original": "https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/gemini_benchmarks_cropped_light2x_1PPmDuP.gif"
    },
    {
      "id": "openai-o3:hle:no-tools-unstated:1",
      "model": "openai-o3",
      "model_name": "OpenAI o3",
      "model_date": "2025-04-16",
      "event": "https://frontierlog.ai/event/2025-04-16-o3-and-o4-mini",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 20.32,
      "unit": "%",
      "printed": "20.32",
      "reported_by": "openai",
      "run_by": null,
      "label": "o3 (no tools)",
      "source": "https://openai.com/index/introducing-o3-and-o4-mini",
      "source_title": "Introducing OpenAI o3 and o4-mini",
      "source_date": "2025-04-16",
      "where": "'What's changed' section, bar chart 'Humanity's Last Exam', 3rd bar 'o3 (no tools)' (page screenshot (job 20879 capture), part 02)",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://blog.google/products-and-platforms/products/gemini/gemini-2-5-pro-latest-preview/",
          "name": "Google post",
          "title": "Try the latest Gemini 2.5 Pro before general availability.",
          "lab": "google",
          "date": "2025-06-05",
          "printed": "20.3%",
          "where": "Benchmark table image (Gemini 2.5 Pro Preview 06-05); row 'Humanity's Last Exam (no tools)'; column 'OpenAI o3 High'. Footnote: non-Gemini from providers' self-reported numbers unless noted; 'For OpenAI models high level of reasoning is shown where results are available (except for GPQA, AIME 2025, SWE-Bench, FACTS, MMMU)'; where provider numbers unavailable, HLE from agi.safe.ai / Scale leaderboard."
        },
        {
          "url": "https://openai.com/index/introducing-chatgpt-agent",
          "name": "OpenAI launches ChatGPT agent",
          "title": "Introducing ChatGPT agent",
          "lab": "openai",
          "date": "2025-07-17",
          "printed": "20.3%",
          "where": "Humanity's Last Exam (Full Set) chart, accuracy pass@1, bar 'OpenAI o3 (no tools)'"
        },
        {
          "url": "https://storage.googleapis.com/deepmind-media/gemini/gemini_v2_5_report.pdf",
          "name": "Gemini 2.5 GA and 2.5 Flash-Lite card",
          "title": "Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities",
          "lab": "google",
          "date": "2025-06-17",
          "printed": "20.3%",
          "where": "Gemini 2.5 technical report, Table 4 (p.14); row 'Humanity's Last Exam, no tools'; column 'o3 high'. Report text: non-Gemini results from providers' self-reported numbers unless mentioned; 'Results for Humanity's Last Exam results are sourced from Scale's leaderboard'. (sha256 of PDF d10cf3e9...)"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/392f30ae76e5713f.png"
    },
    {
      "id": "openai-o3:hle:no-tools-unstated:2",
      "model": "openai-o3",
      "model_name": "OpenAI o3",
      "model_date": "2025-04-16",
      "event": "https://frontierlog.ai/event/2025-04-16-o3-and-o4-mini",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 20.2,
      "unit": "%",
      "printed": "20.2%",
      "reported_by": "openai",
      "run_by": null,
      "label": "OpenAI o3 (high)",
      "source": "https://openai.com/index/introducing-gpt-5-for-developers",
      "source_title": "Introducing GPT-5 for developers",
      "source_date": "2025-08-07",
      "where": "intelligence table (columns GPT-5 (high) ... GPT-4.1 nano), row 'HLE[1] (no tools)', column 'OpenAI o3 (high)'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "high reasoning 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/c0acb8e4e87e723e.png"
    },
    {
      "id": "openai-o4-mini:hle:no-tools-unstated:1",
      "model": "openai-o4-mini",
      "model_name": "o4-mini",
      "model_date": "2025-04-16",
      "event": "https://frontierlog.ai/event/2025-04-16-o3-and-o4-mini",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 14.28,
      "unit": "%",
      "printed": "14.28",
      "reported_by": "openai",
      "run_by": null,
      "label": "o4-mini (no tools)",
      "source": "https://openai.com/index/introducing-o3-and-o4-mini",
      "source_title": "Introducing OpenAI o3 and o4-mini",
      "source_date": "2025-04-16",
      "where": "'What's changed' section, bar chart 'Humanity's Last Exam', 5th bar 'o4-mini (no tools)' (page screenshot (job 20879 capture), part 02)",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-26",
      "corroborated_by": [
        {
          "url": "https://blog.google/products-and-platforms/products/gemini/gemini-2-5-flash-preview/",
          "name": "Gemini 2.5 Flash",
          "title": "Developers can now start building with Gemini 2.5 Flash.",
          "lab": "google",
          "date": "2025-04-17",
          "printed": "14.3%",
          "where": "Benchmark table image, row 'Humanity's Last Exam (no tools)', column 'OpenAI o4-mini'; Google restating (HLE may come from agi.safe.ai/Scale leaderboard where provider numbers are unavailable)"
        },
        {
          "url": "https://blog.google/innovation-and-ai/models-and-research/google-deepmind/google-gemini-updates-io-2025/",
          "name": "Google introduces Deep Think for Gemini 2.5 Pro",
          "title": "Gemini 2.5: Our most intelligent models are getting even better",
          "lab": "google",
          "date": "2025-05-20",
          "printed": "14.3%",
          "where": "Gemini 2.5 Flash (05-20) benchmark table image, row 'Humanity's Last Exam (no tools)', column 'OpenAI o4-mini'; Google restating, same HLE sourcing caveat"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/392f30ae76e5713f.png"
    },
    {
      "id": "gemini-2-5-flash:hle:no-tools-unstated:1",
      "model": "gemini-2-5-flash",
      "model_name": "Gemini 2.5 Flash",
      "model_date": "2025-04-17",
      "event": "https://frontierlog.ai/event/2025-04-17-gemini-2-5-flash",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 12.1,
      "unit": "%",
      "printed": "12.1%",
      "reported_by": "google",
      "run_by": null,
      "label": "Gemini 2.5 Flash Preview (04-17) Thinking",
      "source": "https://blog.google/products-and-platforms/products/gemini/gemini-2-5-flash-preview/",
      "source_title": "Developers can now start building with Gemini 2.5 Flash.",
      "source_date": "2025-04-17",
      "where": "benchmark table (image), row 'Reasoning & knowledge, Humanity's Last Exam (no tools)', column 'Gemini 2.5 Flash Preview (04-17) Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "thinking; pass@1",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://developers.googleblog.com/en/start-building-with-gemini-25-flash/",
          "name": "Gemini 2.5 Flash card",
          "title": "Start building with Gemini 2.5 Flash",
          "lab": "google",
          "date": "2025-04-17",
          "printed": "12.1%",
          "where": "Benchmark table image, row Humanity's Last Exam (no tools), column Gemini 2.5 Flash Preview (04-17) Thinking"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/8cd52ccb5d3627c2.gif",
      "source_image_original": "https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/gemini_2-5_flash_benchmarks_apr.google.gif"
    },
    {
      "id": "gemini-2-5-flash-lite:hle:no-tools-unstated:1",
      "model": "gemini-2-5-flash-lite",
      "model_name": "Gemini 2.5 Flash-Lite",
      "model_date": "2025-06-17",
      "event": "https://frontierlog.ai/event/2025-06-17-gemini-2-5-ga-and-2-5-flash",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 5.1,
      "unit": "%",
      "printed": "5.1%",
      "reported_by": "google",
      "run_by": null,
      "label": "Gemini 2.5 Flash-Lite Preview 06-17 Non-thinking",
      "source": "https://blog.google/products-and-platforms/products/gemini/gemini-2-5-model-family-expands/",
      "source_title": "We’re expanding our Gemini 2.5 family of models",
      "source_date": "2025-06-17",
      "where": "benchmark table (image), row 'Reasoning & knowledge, Humanity's Last Exam (no tools)', column 'Gemini 2.5 Flash-Lite Preview 06-17 Non-thinking'",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "non-thinking; pass@1",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-26",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/4002c58e42db1f64.gif",
      "source_image_original": "https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/gemini_2-5_benchmarks_margin_light2x_1.gif"
    },
    {
      "id": "gemini-2-5-flash-lite:hle:no-tools-unstated:2",
      "model": "gemini-2-5-flash-lite",
      "model_name": "Gemini 2.5 Flash-Lite",
      "model_date": "2025-06-17",
      "event": "https://frontierlog.ai/event/2025-06-17-gemini-2-5-ga-and-2-5-flash",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 6.9,
      "unit": "%",
      "printed": "6.9%",
      "reported_by": "google",
      "run_by": null,
      "label": "Gemini 2.5 Flash-Lite Preview 06-17 Thinking",
      "source": "https://blog.google/products-and-platforms/products/gemini/gemini-2-5-model-family-expands/",
      "source_title": "We’re expanding our Gemini 2.5 family of models",
      "source_date": "2025-06-17",
      "where": "benchmark table (image), row 'Reasoning & knowledge, Humanity's Last Exam (no tools)', column 'Gemini 2.5 Flash-Lite Preview 06-17 Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "thinking; pass@1",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-26",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/4002c58e42db1f64.gif",
      "source_image_original": "https://storage.googleapis.com/gweb-uniblog-publish-prod/original_images/gemini_2-5_benchmarks_margin_light2x_1.gif"
    },
    {
      "id": "gpt-oss-120b:hle:no-tools-unstated:1",
      "model": "gpt-oss-120b",
      "model_name": "gpt-oss-120b",
      "model_date": "2025-08-05",
      "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 14.9,
      "unit": "%",
      "printed": "14.9",
      "reported_by": "openai",
      "run_by": null,
      "label": "gpt-oss-120b (without tools)",
      "source": "https://openai.com/index/introducing-gpt-oss",
      "source_title": "Introducing gpt-oss",
      "source_date": "2025-08-05",
      "where": "'Evaluations' section, bar chart 'Humanity's Last Exam / Expert-level questions across subjects', 2nd bar 'gpt-oss-120b (without tools)' (page screenshot (job 20879 capture), part 02)",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "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/ec5cfc228864ff6c.png"
    },
    {
      "id": "gpt-5-mini:hle:no-tools-unstated:1",
      "model": "gpt-5-mini",
      "model_name": "GPT-5 mini",
      "model_date": "2025-08-07",
      "event": "https://frontierlog.ai/event/2025-08-07-gpt-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 16.7,
      "unit": "%",
      "printed": "16.7%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5 mini (high)",
      "source": "https://openai.com/index/introducing-gpt-5-for-developers",
      "source_title": "Introducing GPT-5 for developers",
      "source_date": "2025-08-07",
      "where": "'Detailed benchmarks' section, 'Intelligence' table (columns GPT-5(high), GPT-5 mini(high), GPT-5 nano(high), ...), row 'HLE[1](no tools)', column 'GPT-5 mini(high)'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-11-06",
      "corroborated_by": [
        {
          "url": "https://x.ai/news/grok-4-fast",
          "name": "xAI post",
          "title": "Grok 4 Fast",
          "lab": "xai",
          "date": "2025-09-19",
          "printed": "16.7 %",
          "where": "\"Benchmark pass@1\" HTML table, row \"HLE (no tools)\", column \"GPT-5 Mini (High)\""
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/4a9ddbdad8a76ac3.png"
    },
    {
      "id": "gpt-5-mini:hle:no-tools-unstated:2",
      "model": "gpt-5-mini",
      "model_name": "GPT-5 mini",
      "model_date": "2025-08-07",
      "event": "https://frontierlog.ai/event/2025-08-07-gpt-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 18.3,
      "unit": "%",
      "printed": "18.3%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5 mini (high)",
      "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": "detailed results, 'Intelligence' table (HTML; columns GPT-5.4 (xhigh), GPT-5.4 mini (xhigh), GPT-5.4 nano (xhigh), GPT-5 mini (high¹)), row 'HLE w/o tools', column 'GPT-5 mini (high¹)'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "high reasoning effort, the highest available",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-07",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/6dd7cf7ff7e77d25.png"
    },
    {
      "id": "gpt-5-nano:hle:no-tools-unstated:1",
      "model": "gpt-5-nano",
      "model_name": "GPT-5 nano",
      "model_date": "2025-08-07",
      "event": "https://frontierlog.ai/event/2025-08-07-gpt-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 8.7,
      "unit": "%",
      "printed": "8.7%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5 nano (high)",
      "source": "https://openai.com/index/introducing-gpt-5-for-developers",
      "source_title": "Introducing GPT-5 for developers",
      "source_date": "2025-08-07",
      "where": "'Detailed benchmarks' section, 'Intelligence' table (columns GPT-5(high), GPT-5 mini(high), GPT-5 nano(high), ...), row 'HLE[1](no tools)', column 'GPT-5 nano(high)'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "high reasoning effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-07",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/4a9ddbdad8a76ac3.png"
    },
    {
      "id": "deepseek-v3-1-terminus:hle:no-tools-unstated:1",
      "model": "deepseek-v3-1-terminus",
      "model_name": "DeepSeek-V3.1-Terminus",
      "model_date": "2025-09-22",
      "event": "https://frontierlog.ai/event/2025-09-22-deepseek-v3-1-terminus",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 21.7,
      "unit": "%",
      "printed": "21.7",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DeepSeek-V3.1-Terminus",
      "source": "https://api-docs.deepseek.com/news/news250922/",
      "source_title": "DeepSeek-V3.1-Terminus",
      "source_date": "2025-09-22",
      "where": "table image, section 'reasoning mode w/o tool use', row \"Humanity's Last Exam\", column 'DeepSeek-V3.1-Terminus'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "reasoning mode; full or text-only set not stated",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Terminus",
          "name": "DeepSeek-V3.1-Terminus card",
          "title": "deepseek-ai/DeepSeek-V3.1-Terminus model card",
          "lab": "deepseek",
          "date": "2025-09-22",
          "printed": "21.7",
          "where": "Table, 'Reasoning Mode w/o Tool Use', row 'Humanity's Last Exam', column 'DeepSeek-V3.1-Terminus'"
        },
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp",
          "name": "DeepSeek-V3.2-Exp card",
          "title": "deepseek-ai/DeepSeek-V3.2-Exp model card",
          "lab": "deepseek",
          "date": "2025-09-29",
          "printed": "21.7",
          "where": "Table, 'Reasoning Mode w/o Tool Use', row 'Humanity's Last Exam', column 'DeepSeek-V3.1-Terminus'"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/93e083b75de124e9.png",
      "source_image_original": "https://api-docs.deepseek.com/zh-cn/img/deepseek_v3.1_terminus_benchmark_en.png"
    },
    {
      "id": "qwen3-vl-235b-a22b-thinking:hle:no-tools-unstated:1",
      "model": "qwen3-vl-235b-a22b-thinking",
      "model_name": "Qwen3-VL-235B-A22B-Thinking",
      "model_date": "2025-09-23",
      "event": "https://frontierlog.ai/event/2025-09-23-qwen3-vl",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 13.6,
      "unit": "%",
      "printed": "13.6",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3-VL 235B-A22B Thinking",
      "source": "https://qwen.ai/blog?id=qwen3-vl",
      "source_title": "Qwen3-VL: Sharper Vision, Deeper Thought, Broader Action",
      "source_date": "2025-09-23",
      "where": "text-benchmark table image for the thinking model (table_thinking_text.jpg), Reasoning block, row 'HLE', column 'Qwen3-VL 235B-A22B Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "thinking model; full or text-only set not stated",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-20",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/4cce76a4c25eaef1.jpg",
      "source_image_original": "https://qianwen-res.oss-accelerate.aliyuncs.com/Qwen3-VL/table_thinking_text.jpg"
    },
    {
      "id": "deepseek-v3-2-exp:hle:no-tools-unstated:1",
      "model": "deepseek-v3-2-exp",
      "model_name": "DeepSeek-V3.2-Exp",
      "model_date": "2025-09-29",
      "event": "https://frontierlog.ai/event/2025-09-29-deepseek-v3-2-exp",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 19.8,
      "unit": "%",
      "printed": "19.8",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DeepSeek-V3.2-Exp",
      "source": "https://api-docs.deepseek.com/news/news250929/",
      "source_title": "Introducing DeepSeek-V3.2-Exp",
      "source_date": "2025-09-29",
      "where": "benchmark table image, section 'General', row \"Humanity's Last Exam\", column 'DeepSeek-V3.2-Exp'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "full or text-only set not stated",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp",
          "name": "DeepSeek-V3.2-Exp card",
          "title": "deepseek-ai/DeepSeek-V3.2-Exp model card",
          "lab": "deepseek",
          "date": "2025-09-29",
          "printed": "19.8",
          "where": "Table, 'Reasoning Mode w/o Tool Use', row 'Humanity's Last Exam', column 'DeepSeek-V3.2-Exp'"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/4d9152a1f18ce8ee.webp",
      "source_image_original": "https://api-docs.deepseek.com/img/v3_2_benchmark.webp"
    },
    {
      "id": "glm-4-6:hle:no-tools-unstated:1",
      "model": "glm-4-6",
      "model_name": "GLM-4.6",
      "model_date": "2025-09-30",
      "event": "https://frontierlog.ai/event/2025-09-30-glm-4-6",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 17.2,
      "unit": "%",
      "printed": "17.2",
      "reported_by": "zai",
      "run_by": null,
      "label": "GLM-4.6",
      "source": "https://z.ai/blog/glm-4.6",
      "source_title": "GLM-4.6: Advanced Agentic, Reasoning and Coding Capabilities",
      "source_date": "2025-09-30",
      "where": "benchmark chart image (eight panels), panel 'HLE', first bar 'GLM-4.6', lower value label inside the bar",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "128K context",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://z.ai/blog/glm-4.7",
          "name": "GLM-4.7",
          "title": "GLM-4.7: Advancing the Coding Capability",
          "lab": "zai",
          "date": "2025-12-22",
          "printed": "17.2",
          "where": "Benchmark table, row 'HLE (separate from 'HLE w/ Tools' row)', column 'GLM-4.6'"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/e2ec892bfba7273b.png",
      "source_image_original": "https://z-cdn.chatglm.cn/z-blog/glm-4-6/coding_benchmark.png"
    },
    {
      "id": "gpt-5-1:hle:no-tools-unstated:1",
      "model": "gpt-5-1",
      "model_name": "GPT-5.1",
      "model_date": "2025-11-12",
      "event": "https://frontierlog.ai/event/2025-11-12-gpt-5-1",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 25.7,
      "unit": "%",
      "printed": "25.7%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.1 Thinking",
      "source": "https://openai.com/index/introducing-gpt-5-2",
      "source_title": "Introducing GPT-5.2",
      "source_date": "2025-12-11",
      "where": "science and math eval table, column 'GPT-5.1 Thinking', row 'HLE (no tools)'",
      "own": false,
      "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://z.ai/blog/glm-4.7",
          "name": "GLM-4.7",
          "title": "GLM-4.7: Advancing the Coding Capability",
          "lab": "zai",
          "date": "2025-12-22",
          "printed": "25.7",
          "where": "Benchmark table, row 'HLE (separate from 'HLE w/ Tools' row)', column 'GPT-5.1 High'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/22817d0b2b0a2432.png"
    },
    {
      "id": "gpt-5-2:hle:no-tools-unstated:1",
      "model": "gpt-5-2",
      "model_name": "GPT-5.2",
      "model_date": "2025-12-11",
      "event": "https://frontierlog.ai/event/2025-12-11-gpt-5-2",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 34.5,
      "unit": "%",
      "printed": "34.5%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.2 Thinking",
      "source": "https://openai.com/index/introducing-gpt-5-2",
      "source_title": "Introducing GPT-5.2",
      "source_date": "2025-12-11",
      "where": "science and math table, row 'HLE (no tools)', column 'GPT-5.2 Thinking'",
      "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/news/claude-opus-4-6",
          "name": "Claude Opus 4.6",
          "title": "Claude Opus 4.6",
          "lab": "anthropic",
          "date": "2026-02-05",
          "printed": "34.5",
          "where": "Bar chart 'Multidisciplinary reasoning, Humanity's Last Exam', bar 'GPT-5.2', lighter 'Without tools' bar"
        },
        {
          "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-1-pro/",
          "name": "Gemini 3.1 Pro",
          "title": "Gemini 3.1 Pro: A smarter model for your most complex tasks",
          "lab": "google",
          "date": "2026-02-19",
          "printed": "34.5%",
          "where": "Benchmark table image; row 'Humanity's Last Exam, No tools'; column gpt-5-2"
        },
        {
          "url": "https://openai.com/index/introducing-gpt-5-4",
          "name": "GPT-5.4",
          "title": "Introducing GPT-5.4",
          "lab": "openai",
          "date": "2026-03-05",
          "printed": "34.5%",
          "where": "table 6 \"Humanity's Last Exam (no tools)\" row, \"GPT-5.2\" column"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/22817d0b2b0a2432.png"
    },
    {
      "id": "nemotron-3-nano-30b-a3b:hle:no-tools-unstated:1",
      "model": "nemotron-3-nano-30b-a3b",
      "model_name": "Nemotron 3 Nano 30B-A3B",
      "model_date": "2025-12-15",
      "event": "https://frontierlog.ai/event/2025-12-15-nemotron-3",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 10.2,
      "unit": "%",
      "printed": "10.2",
      "reported_by": "nvidia",
      "run_by": null,
      "label": "NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",
      "source": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8/blob/fa5699ff0bb9da723217af52488c0f9653512ffb/README.md",
      "source_title": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8 model card (launch-day revision)",
      "source_date": "2025-12-15",
      "where": "'Reasoning Benchmark Evaluations' table, row 'HLE (no tools)', column 'NVIDIA-Nemotron-3-Nano-30B-A3B-BF16'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4/blob/6efb4a2a1c1fa277ce7b3df7a1416255011b1c99/README.md",
          "name": "Nemotron 3 card",
          "title": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 model card (revision of 2026-08-24)",
          "lab": "nvidia",
          "date": "2025-12-15",
          "printed": "10.2",
          "where": "accuracy table (BF16 | FP8 | NVFP4), first (BF16) column, HLE (no tools) row"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/4ff9ee8b68e61bc2.png"
    },
    {
      "id": "nemotron-3-nano-30b-a3b:hle:no-tools-unstated:2",
      "model": "nemotron-3-nano-30b-a3b",
      "model_name": "Nemotron 3 Nano 30B-A3B",
      "model_date": "2025-12-15",
      "event": "https://frontierlog.ai/event/2025-12-15-nemotron-3",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 10.57,
      "unit": "%",
      "printed": "10.57",
      "reported_by": "nvidia",
      "run_by": null,
      "label": "N-3-Nano",
      "source": "https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Nano-Technical-Report.pdf",
      "source_title": "Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning",
      "source_date": "2025-12-15",
      "where": "Table 3 (PDF page 22), row 'HLE (no tools)', column 'N-3-Nano'",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16/blob/bf77c3174f68ad409e1c2aa60daeb46e32d1c606/README.md",
          "name": "Nemotron 3 card",
          "title": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 model card (revision of 2026-08-24)",
          "lab": "nvidia",
          "date": "2025-12-15",
          "printed": "10.6",
          "where": "Benchmarks table, 'Nemotron-3-Nano' (first, BF16) column, HLE (no tools) row"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/97664fca11ce8e21.png"
    },
    {
      "id": "glm-4-7:hle:no-tools-unstated:1",
      "model": "glm-4-7",
      "model_name": "GLM-4.7",
      "model_date": "2025-12-22",
      "event": "https://frontierlog.ai/event/2025-12-22-glm-4-7",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 24.8,
      "unit": "%",
      "printed": "24.8",
      "reported_by": "zai",
      "run_by": null,
      "label": "GLM-4.7",
      "source": "https://z.ai/blog/glm-4.7",
      "source_title": "GLM-4.7: Advancing the Coding Capability",
      "source_date": "2025-12-22",
      "where": "benchmark table, row 'HLE', column 'GLM-4.7'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://z.ai/blog/glm-5",
          "name": "GLM-5",
          "title": "GLM-5: From Vibe Coding to Agentic Engineering",
          "lab": "zai",
          "date": "2026-02-12",
          "printed": "24.8",
          "where": "Chart 'LLM Performance Evaluation: Agentic, Reasoning and Coding', panel 'Humanity's Last Exam', bar 'GLM-4.7' (lower label, not marked w/ Tools)"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/b52f90e840f8cbd6.png"
    },
    {
      "id": "minimax-m2-1:hle:no-tools-unstated:1",
      "model": "minimax-m2-1",
      "model_name": "MiniMax M2.1",
      "model_date": "2025-12-23",
      "event": "https://frontierlog.ai/event/2025-12-23-minimax-m2-1",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 22.2,
      "unit": "%",
      "printed": "22.2",
      "reported_by": "minimax",
      "run_by": null,
      "label": "MiniMax-M2.1",
      "source": "https://www.minimax.io/news/minimax-m25",
      "source_title": "MiniMax M2.5: Built for Real-World Productivity.",
      "source_date": "2026-02-12",
      "where": "later report: MiniMax M2.5 post, Appendix 'Further benchmark results' table image, row 'HLE w/o tools', column 'MiniMax-M2.1'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "in-house run with Artificial Analysis Intelligence Index methods",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/MiniMaxAI/MiniMax-M2.1",
          "name": "MiniMax M2.1 card",
          "title": "MiniMaxAI/MiniMax-M2.1 model card",
          "lab": "minimax",
          "date": "2025-12-23",
          "printed": "22.2",
          "where": "third markdown table, row 'HLE w/o tools', column MiniMax-M2.1"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/5af8b315069f9686.png",
      "source_image_original": "https://file.cdn.minimax.io/public/8c019213-b0d5-4ee8-9273-6d9b799abeae.png"
    },
    {
      "id": "qwen3-max-thinking:hle:no-tools-unstated:1",
      "model": "qwen3-max-thinking",
      "model_name": "Qwen3-Max-Thinking",
      "model_date": "2026-01-26",
      "event": "https://frontierlog.ai/event/2026-01-26-qwen3-max-thinking",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 36.5,
      "unit": "%",
      "printed": "36.5",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3-Max-Thinking (with TTS)",
      "source": "https://qwen.ai/blog?id=qwen3-max-thinking",
      "source_title": "Pushing Qwen3-Max-Thinking Beyond its Limits",
      "source_date": "2026-01-26",
      "where": "score chart image (seven panels, below the banner), panel 'Humanity's Last Exam', first bar, upper value label (legend 'Qwen3-Max-Thinking (with TTS)')",
      "own": true,
      "primary": false,
      "parallel": true,
      "notes": "heavy mode: test-time scaling over parallel trajectories",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/80d8b6e334ec18d8.png",
      "source_image_original": "https://qianwen-res.oss-accelerate-overseas.aliyuncs.com/Qwen3-Max-Thinking/score.png"
    },
    {
      "id": "minimax-m2-5:hle:no-tools-unstated:1",
      "model": "minimax-m2-5",
      "model_name": "MiniMax M2.5",
      "model_date": "2026-02-12",
      "event": "https://frontierlog.ai/event/2026-02-12-minimax-m2-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 19.4,
      "unit": "%",
      "printed": "19.4",
      "reported_by": "minimax",
      "run_by": null,
      "label": "MiniMax-M2.5",
      "source": "https://www.minimax.io/news/minimax-m25",
      "source_title": "MiniMax M2.5: Built for Real-World Productivity.",
      "source_date": "2026-02-12",
      "where": "Appendix 'Further benchmark results' table image, row 'HLE w/o tools', column 'MiniMax-M2.5'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "in-house run with Artificial Analysis Intelligence Index methods; full or text-only set not stated",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-20",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/5af8b315069f9686.png",
      "source_image_original": "https://file.cdn.minimax.io/public/8c019213-b0d5-4ee8-9273-6d9b799abeae.png"
    },
    {
      "id": "qwen3-5-397b-a17b:hle:no-tools-unstated:1",
      "model": "qwen3-5-397b-a17b",
      "model_name": "Qwen3.5-397B-A17B",
      "model_date": "2026-02-16",
      "event": "https://frontierlog.ai/event/2026-02-16-qwen3-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 28.7,
      "unit": "%",
      "printed": "28.7",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.5-397B-A17B",
      "source": "https://qwen.ai/blog?id=qwen3.5",
      "source_title": "Qwen3.5: Towards Native Multimodal Agents",
      "source_date": "2026-02-16",
      "where": "'Performance' section, 'Language' table, row 'HLE', column 'Qwen3.5-397B-A17B'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/a2e98da83152d308.png"
    },
    {
      "id": "gpt-5-4:hle:no-tools-unstated: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-unstated",
      "setting_name": "No tools, set not stated",
      "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": "nemotron-3-super-120b-a12b:hle:no-tools-unstated:1",
      "model": "nemotron-3-super-120b-a12b",
      "model_name": "Nemotron 3 Super 120B-A12B",
      "model_date": "2026-03-11",
      "event": "https://frontierlog.ai/event/2026-03-11-nemotron-3-super",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 18.26,
      "unit": "%",
      "printed": "18.26",
      "reported_by": "nvidia",
      "run_by": null,
      "label": "Nemotron-3-Super",
      "source": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8/blob/1408abd63ba6ba1c37a78b94d332dac35da45370/README.md",
      "source_title": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8 model card (launch-day revision)",
      "source_date": "2026-03-11",
      "where": "benchmark table under the accuracy chart, row 'HLE (no tools)', column 'Nemotron-3-Super'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Super-Technical-Report.pdf",
          "name": "Nemotron 3 Super card",
          "title": "Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning",
          "lab": "nvidia",
          "date": "2026-03-11",
          "printed": "18.26",
          "where": "Table 5, HLE (no tools) row, N-3-Super column"
        },
        {
          "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16/blob/2dc98e2afe4face0e4ce40972a915c45368bd34a/README.md",
          "name": "Nemotron 3 Super card",
          "title": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 model card (revision of 2026-08-25)",
          "lab": "nvidia",
          "date": "2026-03-11",
          "printed": "18.26",
          "where": "Benchmarks table, 'Nemotron 3 Super' column, HLE (no tools) row"
        },
        {
          "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4/blob/ff433f5493e25d631c9f12b5d55c674229923d02/README.md",
          "name": "Nemotron 3 Super card",
          "title": "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4 model card (revision of 2026-08-24)",
          "lab": "nvidia",
          "date": "2026-03-11",
          "printed": "18.26",
          "where": "accuracy table, 'Nemotron-3-Super' (BF16) column, HLE (no tools) row"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/0a1e755f970190a7.png"
    },
    {
      "id": "gpt-5-4-mini:hle:no-tools-unstated:1",
      "model": "gpt-5-4-mini",
      "model_name": "GPT-5.4 mini",
      "model_date": "2026-03-17",
      "event": "https://frontierlog.ai/event/2026-03-17-gpt-5-4-mini-nano",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 28.2,
      "unit": "%",
      "printed": "28.2%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.4 mini (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), ...), row 'HLE w/o tools', column 'GPT-5.4 mini (xhigh)'",
      "own": true,
      "primary": true,
      "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"
    },
    {
      "id": "gpt-5-4-nano:hle:no-tools-unstated:1",
      "model": "gpt-5-4-nano",
      "model_name": "GPT-5.4 nano",
      "model_date": "2026-03-17",
      "event": "https://frontierlog.ai/event/2026-03-17-gpt-5-4-mini-nano",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 24.3,
      "unit": "%",
      "printed": "24.3%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT-5.4 nano (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), ...), row 'HLE w/o tools', column 'GPT-5.4 nano (xhigh)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-26",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/ec1c7aef5bba5db5.png"
    },
    {
      "id": "qwen3-6-plus:hle:no-tools-unstated:1",
      "model": "qwen3-6-plus",
      "model_name": "Qwen3.6-Plus",
      "model_date": "2026-04-02",
      "event": "https://frontierlog.ai/event/2026-04-02-qwen3-6-plus",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 28.8,
      "unit": "%",
      "printed": "28.8",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.6-Plus",
      "source": "https://qwen.ai/blog?id=qwen3.6",
      "source_title": "Qwen3.6-Plus: Towards Real World Agents",
      "source_date": "2026-04-02",
      "where": "'Performance' section, 'Language' table, row 'HLE', column 'Qwen3.6-Plus'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/6f4855b376c670b4.png"
    },
    {
      "id": "qwen3-6-35b-a3b:hle:no-tools-unstated:1",
      "model": "qwen3-6-35b-a3b",
      "model_name": "Qwen3.6-35B-A3B",
      "model_date": "2026-04-15",
      "event": "https://frontierlog.ai/event/2026-04-15-qwen3-6-35b-a3b",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 21.4,
      "unit": "%",
      "printed": "21.4",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.6-35BA3B",
      "source": "https://qwen.ai/blog?id=qwen3.6-35b-a3b",
      "source_title": "Qwen3.6-35B-A3B: Agentic Coding Power, Now Open to All",
      "source_date": "2026-04-15",
      "where": "Language table (first table), STEM & Reasoning block, row 'HLE', column 'Qwen3.6-35BA3B'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "full or text-only set not stated",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/Qwen/Qwen3.6-35B-A3B",
          "name": "Qwen3.6-35B-A3B card",
          "title": "Qwen/Qwen3.6-35B-A3B model card",
          "lab": "alibaba",
          "date": "2026-04-15",
          "printed": "21.4",
          "where": "HF model card, HTML table, row HLE, column Qwen3.6-35BA3B"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/29389a297e0321ef.png"
    },
    {
      "id": "gpt-5-5:hle:no-tools-unstated:1",
      "model": "gpt-5-5",
      "model_name": "GPT-5.5",
      "model_date": "2026-04-23",
      "event": "https://frontierlog.ai/event/2026-04-23-gpt-5-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 41.4,
      "unit": "%",
      "printed": "41.4%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.5",
      "source": "https://openai.com/index/introducing-gpt-5-5",
      "source_title": "Introducing GPT-5.5",
      "source_date": "2026-04-23",
      "where": "academic eval table, column 'GPT‑5.5', row 'Humanity's Last Exam (no 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-fable-5-mythos-5-system-card",
          "name": "Claude Fable 5 and Mythos 5 card",
          "title": "Claude Fable 5 and Claude Mythos 5 System Card",
          "lab": "anthropic",
          "date": "2026-06-09",
          "printed": "41.4",
          "where": "Table 8.1.A (p.251), row Humanity's Last Exam / No tools, column GPT-5.5"
        },
        {
          "url": "https://www.anthropic.com/claude-opus-4-8-system-card",
          "name": "Claude Opus 4.8 card",
          "title": "Claude Opus 4.8 System Card",
          "lab": "anthropic",
          "date": "2026-05-28",
          "printed": "41.4",
          "where": "Table 8.1.A (p.194), row Humanity's Last Exam / No tools, column GPT-5.5"
        },
        {
          "url": "https://www.anthropic.com/claude-sonnet-5-system-card",
          "name": "Claude Sonnet 5 card",
          "title": "Claude Sonnet 5 System Card",
          "lab": "anthropic",
          "date": "2026-06-30",
          "printed": "41.4",
          "where": "Table 8.1.A (p.115), row Humanity's Last Exam / No tools, column GPT-5.5"
        },
        {
          "url": "https://www.anthropic.com/news/claude-opus-4-8",
          "name": "Claude Opus 4.8",
          "title": "Introducing Claude Opus 4.8",
          "lab": "anthropic",
          "date": "2026-05-28",
          "printed": "41.4%",
          "where": "benchmark table image, row Humanity's Last Exam (no tools), column GPT-5.5"
        },
        {
          "url": "https://www.kimi.ai/blog/kimi-k3",
          "name": "Kimi K3",
          "title": "Kimi K3: Open Frontier Intelligence",
          "lab": "moonshot",
          "date": "2026-07-16",
          "printed": "41.4",
          "where": "Full Benchmark Table, row HLE-Full ; column GPT 5.5 (xhigh)"
        },
        {
          "url": "https://z.ai/blog/glm-5.2",
          "name": "GLM-5.2",
          "title": "GLM-5.2: Built for Long-Horizon Tasks",
          "lab": "zai",
          "date": "2026-06-16",
          "printed": "41.4",
          "where": "Chart 'LLM Performance Evaluation' (all models at max thinking effort), panel 'Humanity's Last Exam', bar 'GPT-5.5' (lower label, not marked w/ Tools)"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/4fc1be3b62dac325.png"
    },
    {
      "id": "deepseek-v4-pro:hle:no-tools-unstated:1",
      "model": "deepseek-v4-pro",
      "model_name": "DeepSeek-V4-Pro",
      "model_date": "2026-04-24",
      "event": "https://frontierlog.ai/event/2026-04-24-deepseek-v4-preview",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 37.7,
      "unit": "%",
      "printed": "37.7",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DS-V4-Pro Max",
      "source": "https://api-docs.deepseek.com/news/news260424/",
      "source_title": "DeepSeek V4 Preview Release",
      "source_date": "2026-04-24",
      "where": "benchmark table image (v4-benchmark-2), section 'Knowledge & Reasoning', row 'HLE (Pass@1)', column 'DS-V4-Pro Max'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "max reasoning effort; full or text-only set not stated",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://api-docs.deepseek.com/news/news260813/",
          "name": "DeepSeek-V4-Pro GA",
          "title": "DeepSeek-V4-Pro GA Release",
          "lab": "deepseek",
          "date": "2026-08-13",
          "printed": "37.7",
          "where": "Benchmark table image, row 'HLE (wo/w tools)', column 'DeepSeek-V4-Pro-Preview', printed '37.7/48.2'"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/9226203de8e24b33.png",
      "source_image_original": "https://api-docs.deepseek.com/img/v4-benchmark-2.png"
    },
    {
      "id": "qwen3-7-max:hle:no-tools-unstated:1",
      "model": "qwen3-7-max",
      "model_name": "Qwen3.7-Max",
      "model_date": "2026-05-20",
      "event": "https://frontierlog.ai/event/2026-05-20-qwen3-7-max",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 41.4,
      "unit": "%",
      "printed": "41.4",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.7-Max",
      "source": "https://qwen.ai/blog?id=qwen3.7",
      "source_title": "Qwen3.7: The Agent Frontier",
      "source_date": "2026-05-20",
      "where": "'Performance' table, row 'HLE', column 'Qwen3.7-Max'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh reasoning effort",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/9d2174c2d8ef254a.png"
    },
    {
      "id": "qwen3-7-plus:hle:no-tools-unstated:1",
      "model": "qwen3-7-plus",
      "model_name": "Qwen3.7-Plus",
      "model_date": "2026-06-01",
      "event": "https://frontierlog.ai/event/2026-06-01-qwen3-7-plus",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 34.7,
      "unit": "%",
      "printed": "34.7",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.7-Plus",
      "source": "https://qwen.ai/blog?id=qwen3.7-plus",
      "source_title": "Qwen3.7-Plus: Multimodal Agent Intelligence",
      "source_date": "2026-06-01",
      "where": "language benchmark table (first table), STEM & Reasoning block, row 'HLE', column 'Qwen3.7-Plus'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "xhigh reasoning effort; full or text-only set not stated",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
          "name": "Qwen3.8-Flash-Next card",
          "title": "Qwen/Qwen3.8-Flash-Next model card",
          "lab": "alibaba",
          "date": "2026-08-26",
          "printed": "34.7",
          "where": "Qwen3.8-Flash-Next HF model card, HTML table, row Multidisciplinary reasoning / HLE, column Qwen3.7-Plus; footnote: HLE judged by GPT-4o"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/0b4fd6fbae7120cb.png"
    },
    {
      "id": "nemotron-3-ultra-550b-a55b:hle:no-tools-unstated:1",
      "model": "nemotron-3-ultra-550b-a55b",
      "model_name": "Nemotron 3 Ultra 550B-A55B",
      "model_date": "2026-06-04",
      "event": "https://frontierlog.ai/event/2026-06-04-nemotron-3-ultra",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 26.7,
      "unit": "%",
      "printed": "26.7",
      "reported_by": "nvidia",
      "run_by": null,
      "label": "Nemotron 3 Ultra BF16",
      "source": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4/blob/02c7d9e60f328a4fb9d990d6b613057b19ab8a8f/README.md",
      "source_title": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 model card (launch-day revision)",
      "source_date": "2026-06-04",
      "where": "benchmark table, row 'HLE (no tools)', column 'Nemotron 3 Ultra BF16'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Ultra-Technical-Report.pdf",
          "name": "Nemotron 3 Ultra card",
          "title": "Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning",
          "lab": "nvidia",
          "date": "2026-06-04",
          "printed": "26.7",
          "where": "Table 10, HLE (no tools) row, N-3-Ultra column"
        },
        {
          "url": "https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16/blob/77df655d5e9f8362164ed14dd8b48f8bce657498/README.md",
          "name": "Nemotron 3 Ultra card",
          "title": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 model card (revision of 2026-08-24)",
          "lab": "nvidia",
          "date": "2026-06-04",
          "printed": "26.7",
          "where": "Benchmarks table, 'N-3-Ultra 550B-A55B' column, HLE (no tools) row"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/e4e42f413aeecae9.png"
    },
    {
      "id": "qwen3-8-max:hle:no-tools-unstated:1",
      "model": "qwen3-8-max",
      "model_name": "Qwen3.8-Max",
      "model_date": "2026-08-03",
      "event": "https://frontierlog.ai/event/2026-08-03-qwen3-8-max",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 43.6,
      "unit": "%",
      "printed": "43.6",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.8-Max",
      "source": "https://qwen.ai/blog?id=qwen3.8",
      "source_title": "Qwen3.8-Max: A New Bar for Coding and Cowork",
      "source_date": "2026-08-03",
      "where": "'Full Benchmark Table' (first table), row 'HLE' under 'General Capabilities', column 'Qwen3.8-Max' (read via the qwen.ai article API, article 65dfab20-97fb-4932-97cd-4f7da4ab51f5)",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "full or text-only set not stated",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-17",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/10d02db2e5d51286.png"
    },
    {
      "id": "qwen3-8-flash-next:hle:no-tools-unstated:1",
      "model": "qwen3-8-flash-next",
      "model_name": "Qwen3.8-Flash-Next",
      "model_date": "2026-08-26",
      "event": "https://frontierlog.ai/event/2026-08-26-qwen3-8-flash-next",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 35.9,
      "unit": "%",
      "printed": "35.9",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.8-Flash-Next",
      "source": "https://qwen.ai/blog?id=qwen3.8-flash-next",
      "source_title": "Qwen3.8-Flash-Next: A New Architecture, Towards Ultimate Cost-Efficiency",
      "source_date": "2026-08-26",
      "where": "Language table (first table), row 'Multidisciplinary reasoning / HLE', column 'Qwen3.8-Flash-Next'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "GPT-4o judge; full or text-only set not stated",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/Qwen/Qwen3.8-Flash-Next",
          "name": "Qwen3.8-Flash-Next card",
          "title": "Qwen/Qwen3.8-Flash-Next model card",
          "lab": "alibaba",
          "date": "2026-08-26",
          "printed": "35.9",
          "where": "README Language table, row 'Multidisciplinary reasoning / HLE', column 'Qwen3.8-Flash-Next'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/ce5ffcfe0edfb313.png"
    },
    {
      "id": "deepseek-v4-1-flash:hle:no-tools-unstated:1",
      "model": "deepseek-v4-1-flash",
      "model_name": "DeepSeek-V4.1-Flash",
      "model_date": "2026-09-10",
      "event": "https://frontierlog.ai/event/2026-09-10-deepseek-v4-1-flash",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 36.8,
      "unit": "%",
      "printed": "36.8",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DeepSeek V4.1-Flash",
      "source": "https://api-docs.deepseek.com/news/news260910/",
      "source_title": "DeepSeek-V4.1-Flash: Smarter, Faster, More Efficient",
      "source_date": "2026-09-10",
      "where": "benchmark table image (second image), row 'HLE', column 'DeepSeek V4.1-Flash', value outside the brackets",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash",
          "name": "DeepSeek-V4.1-Flash card",
          "title": "deepseek-ai/DeepSeek-V4.1-Flash model card",
          "lab": "deepseek",
          "date": "2026-09-10",
          "printed": "36.8",
          "where": "'Comparison with frontier models (Max reasoning effort)' table, row 'HLE (Pass@1)', column 'DS-V4.1-Flash', printed '36.8 (39.1†)'"
        },
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash/blob/main/DeepSeek_V41_Tech_Report.pdf",
          "name": "DeepSeek-V4.1-Flash card",
          "title": "DeepSeek-V4.1-Flash: Pushing the Limits of KV Cache Compression",
          "lab": "deepseek",
          "date": "2026-09-10",
          "printed": "36.8",
          "where": "Instruct results table (Max reasoning effort), row 'HLE (Pass@1)', column 'DS-V4.1-Flash', printed '36.8 (39.1†)'"
        },
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
          "name": "DeepSeek-V4.1-Flash-1017 card",
          "title": "deepseek-ai/DeepSeek-V4.1-Flash-1017 model card",
          "lab": "deepseek",
          "date": "2026-10-17",
          "printed": "36.8",
          "where": "Table, row 'HLE w/o tools', column 'DS-V4.1-Flash', printed '36.8 (39.1†)'"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/19d3b83b1b9e5f5b.png",
      "source_image_original": "https://api-docs.deepseek.com/img/v4.1_260910_benchmark_table_en.png"
    },
    {
      "id": "qwen3-8-omni-flash:hle:no-tools-unstated:1",
      "model": "qwen3-8-omni-flash",
      "model_name": "Qwen3.8-Omni-Flash",
      "model_date": "2026-09-18",
      "event": "https://frontierlog.ai/event/2026-09-18-qwen3-8-omni-flash",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 36.5,
      "unit": "%",
      "printed": "36.5",
      "reported_by": "alibaba",
      "run_by": null,
      "label": "Qwen3.8-Omni-Flash",
      "source": "https://qwen.ai/blog?id=qwen3.8-omni-flash",
      "source_title": "Qwen3.8-Omni-Flash: Omni Senses. Agentic Delivery.",
      "source_date": "2026-09-18",
      "where": "text benchmark table (General Text Capabilities block), row 'HLE', column 'Qwen3.8-Omni-Flash'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "GPT-4o judge; full or text-only set not stated",
      "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/404338e9a04ccbae.png"
    },
    {
      "id": "deepseek-v4-1-flash-1017:hle:no-tools-unstated:1",
      "model": "deepseek-v4-1-flash-1017",
      "model_name": "DeepSeek-V4.1-Flash-1017",
      "model_date": "2026-10-17",
      "event": "https://frontierlog.ai/event/2026-10-17-deepseek-v4-1-flash-1017",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "no-tools-unstated",
      "setting_name": "No tools, set not stated",
      "score": 42.6,
      "unit": "%",
      "printed": "42.6",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DeepSeek V4.1-Flash-1017",
      "source": "https://api-docs.deepseek.com/news/news261017/",
      "source_title": "DeepSeek-V4.1-Flash-1017: Wish-Driven Programming Made Real",
      "source_date": "2026-10-17",
      "where": "benchmark table image (first image), row 'HLE (w/o tools)', column 'DeepSeek V4.1-Flash-1017', first of the two values",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/deepseek-ai/DeepSeek-V4.1-Flash-1017",
          "name": "DeepSeek-V4.1-Flash-1017 card",
          "title": "deepseek-ai/DeepSeek-V4.1-Flash-1017 model card",
          "lab": "deepseek",
          "date": "2026-10-17",
          "printed": "42.6",
          "where": "Table, row 'HLE w/o tools', column 'DS-V4.1-Flash-1017', printed '42.6 (44.0†)'"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/52c22094008778b4.png",
      "source_image_original": "https://api-docs.deepseek.com/img/v4.1_261017_benchmark_table_en.png"
    }
  ]
}