{
  "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": 15,
  "results": [
    {
      "id": "deepseek-r1-0528:hle:with-tools-text-only:1",
      "model": "deepseek-r1-0528",
      "model_name": "DeepSeek-R1-0528",
      "model_date": "2025-05-28",
      "event": "https://frontierlog.ai/event/2025-05-28-deepseek-r1-0528",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 24.8,
      "unit": "%",
      "printed": "24.8",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DeepSeek-R1-0528",
      "source": "https://api-docs.deepseek.com/news/news250821/",
      "source_title": "DeepSeek-V3.1 Release",
      "source_date": "2025-08-21",
      "where": "later report: DeepSeek-V3.1 post, second table image (search agent benchmarks), row 'HLE', column 'DeepSeek-R1-0528'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "search agent",
      "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",
          "name": "DeepSeek-V3.1 card",
          "title": "deepseek-ai/DeepSeek-V3.1 model card",
          "lab": "deepseek",
          "date": "2025-08-21",
          "printed": "24.8",
          "where": "Evaluation table, Search Agent, row 'Humanity's Last Exam (Python + Search)', column 'DeepSeek R1 0528' (text-only subset)"
        }
      ],
      "source_image": "https://frontierlog.ai/sources/c164a27411cd68bc.webp",
      "source_image_original": "https://cdn.deepseek.com/api-docs/v3.1_benchmark_2.webp"
    },
    {
      "id": "deepseek-v3-1:hle:with-tools-text-only:1",
      "model": "deepseek-v3-1",
      "model_name": "DeepSeek-V3.1",
      "model_date": "2025-08-21",
      "event": "https://frontierlog.ai/event/2025-08-21-deepseek-v3-1",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 29.8,
      "unit": "%",
      "printed": "29.8",
      "reported_by": "deepseek",
      "run_by": null,
      "label": "DeepSeek-V3.1",
      "source": "https://api-docs.deepseek.com/news/news250821/",
      "source_title": "DeepSeek-V3.1 Release",
      "source_date": "2025-08-21",
      "where": "second table image (search agent), row 'HLE', column 'DeepSeek-V3.1' (shaded)",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "search agent",
      "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/c164a27411cd68bc.webp",
      "source_image_original": "https://cdn.deepseek.com/api-docs/v3.1_benchmark_2.webp"
    },
    {
      "id": "kimi-k2-thinking:hle:with-tools-text-only:1",
      "model": "kimi-k2-thinking",
      "model_name": "Kimi K2 Thinking",
      "model_date": "2025-11-06",
      "event": "https://frontierlog.ai/event/2025-11-06-kimi-k2-thinking",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 44.9,
      "unit": "%",
      "printed": "44.9",
      "reported_by": "moonshot",
      "run_by": null,
      "label": "K2 Thinking",
      "source": "https://www.kimi.ai/blog/kimi-k2-thinking",
      "source_title": "Introducing Kimi K2 Thinking",
      "source_date": "2025-11-06",
      "where": "'Full Evaluations' table, row 'Humanity's Last Exam (Text-only)', sub-row 'w/ tools [4]', column 'K2 Thinking'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; search, code and browsing tools; up to 120 steps; Hugging Face blocked; INT4 precision",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/zai-org/GLM-4.7",
          "name": "GLM-4.7 card",
          "title": "zai-org/GLM-4.7 model card",
          "lab": "zai",
          "date": "2025-12-22",
          "printed": "44.9",
          "where": "README benchmark table, row 'HLE (w/ Tools)', column 'Kimi K2 Thinking'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/4e2626572b0cef3e.png"
    },
    {
      "id": "kimi-k2-thinking:hle:with-tools-text-only:2",
      "model": "kimi-k2-thinking",
      "model_name": "Kimi K2 Thinking",
      "model_date": "2025-11-06",
      "event": "https://frontierlog.ai/event/2025-11-06-kimi-k2-thinking",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 51.3,
      "unit": "%",
      "printed": "51.3",
      "reported_by": "moonshot",
      "run_by": null,
      "label": "K2 Thinking",
      "source": "https://www.kimi.ai/blog/kimi-k2-thinking",
      "source_title": "Introducing Kimi K2 Thinking",
      "source_date": "2025-11-06",
      "where": "Footnotes, 'For HLE (w/ tools) and the agentic-search benchmarks', item f (score without blocking Hugging Face)",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "Hugging Face access not blocked, which may leak answers; text-only subset",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-20",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "kimi-k2-thinking:hle:with-tools-text-only:3",
      "model": "kimi-k2-thinking",
      "model_name": "Kimi K2 Thinking",
      "model_date": "2025-11-06",
      "event": "https://frontierlog.ai/event/2025-11-06-kimi-k2-thinking",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 51.0,
      "unit": "%",
      "printed": "51.0",
      "reported_by": "moonshot",
      "run_by": null,
      "label": "K2 Thinking",
      "source": "https://www.kimi.ai/blog/kimi-k2-thinking",
      "source_title": "Introducing Kimi K2 Thinking",
      "source_date": "2025-11-06",
      "where": "'Full Evaluations' table, row 'Humanity's Last Exam (Text-only)', sub-row 'heavy [6]', column 'K2 Thinking'",
      "own": true,
      "primary": false,
      "parallel": true,
      "notes": "heavy mode: eight parallel trajectories aggregated into one answer; text-only subset",
      "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/4e2626572b0cef3e.png"
    },
    {
      "id": "glm-5:hle:with-tools-text-only:1",
      "model": "glm-5",
      "model_name": "GLM-5",
      "model_date": "2026-02-12",
      "event": "https://frontierlog.ai/event/2026-02-12-glm-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 50.4,
      "unit": "%",
      "printed": "50.4",
      "reported_by": "zai",
      "run_by": null,
      "label": "GLM-5",
      "source": "https://z.ai/blog/glm-5",
      "source_title": "GLM-5: From Vibe Coding to Agentic Engineering",
      "source_date": "2026-02-12",
      "where": "benchmark table, row 'Humanity's Last Exam w/ Tools', column 'GLM-5 (Thinking)'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; GPT-5.2 judge; 202,752-token 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-5.1",
          "name": "GLM-5.1",
          "title": "GLM-5.1: Towards Long-Horizon Tasks",
          "lab": "zai",
          "date": "2026-04-07",
          "printed": "50.4",
          "where": "Benchmark table, row 'HLE w/ Tools (* = full set for Opus 4.6, Gemini 3.1 Pro, GPT-5.4)', column 'GLM-5'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/8709485a935545d0.png"
    },
    {
      "id": "glm-5-1:hle:with-tools-text-only:1",
      "model": "glm-5-1",
      "model_name": "GLM-5.1",
      "model_date": "2026-04-07",
      "event": "https://frontierlog.ai/event/2026-04-07-glm-5-1",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 52.3,
      "unit": "%",
      "printed": "52.3",
      "reported_by": "zai",
      "run_by": null,
      "label": "GLM-5.1",
      "source": "https://z.ai/blog/glm-5.1",
      "source_title": "GLM-5.1: Towards Long-Horizon Tasks",
      "source_date": "2026-04-07",
      "where": "benchmark table near the end, row 'HLE w/ Tools', column 'GLM-5.1'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; GPT-5.2 judge; 202,752-token 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-5.2",
          "name": "GLM-5.2",
          "title": "GLM-5.2: Built for Long-Horizon Tasks",
          "lab": "zai",
          "date": "2026-06-16",
          "printed": "52.3",
          "where": "Full Benchmark Table, row 'HLE w/ Tools (* = full set for Opus 4.8, GPT-5.5, Gemini 3.1 Pro)', column 'GLM-5.1'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/35c66845b36400f8.png"
    },
    {
      "id": "glm-5-2:hle:with-tools-text-only:1",
      "model": "glm-5-2",
      "model_name": "GLM-5.2",
      "model_date": "2026-06-16",
      "event": "https://frontierlog.ai/event/2026-06-16-glm-5-2",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 54.7,
      "unit": "%",
      "printed": "54.7",
      "reported_by": "zai",
      "run_by": null,
      "label": "GLM-5.2",
      "source": "https://z.ai/blog/glm-5.2",
      "source_title": "GLM-5.2: Built for Long-Horizon Tasks",
      "source_date": "2026-06-16",
      "where": "'Full Benchmark Table', row 'HLE w/ Tools', column 'GLM-5.2'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; 300,000-token context, no context management",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-10-19",
      "check": "human",
      "corroborated": true,
      "corroborated_on": "2026-10-19",
      "corroborated_by": [
        {
          "url": "https://thinkingmachines.ai/news/introducing-inkling/",
          "name": "Inkling",
          "title": "Inkling: Our Open-Weights Model",
          "lab": "thinking-machines",
          "date": "2026-07-15",
          "printed": "54.7%",
          "where": "'Benchmarking Inkling' main table, row 'HLE with tools', column 'GLM 5.2'"
        },
        {
          "url": "https://z.ai/blog/glm-5.3",
          "name": "GLM-5.3",
          "title": "GLM-5.3: Frontier Coding with Emergent Cyber Capabilities",
          "lab": "zai",
          "date": "2026-08-14",
          "printed": "54.7",
          "where": "'Performance across comparison models' table, row 'HLE w/ Tools', column 'GLM-5.2'"
        }
      ],
      "source_capture": "https://frontierlog.ai/captures/112d3aba81edfc46.png"
    },
    {
      "id": "hy3:hle:with-tools-text-only:1",
      "model": "hy3",
      "model_name": "Hy3",
      "model_date": "2026-07-06",
      "event": "https://frontierlog.ai/event/2026-07-06-hy3",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 53.2,
      "unit": "%",
      "printed": "53.2",
      "reported_by": "tencent",
      "run_by": null,
      "label": "Hy3",
      "source": "https://hy.tencent.ai/research/hy3",
      "source_title": "Introducing Hy3",
      "source_date": "2026-07-06",
      "where": "'Appendix: Benchmark Results' table image, row 'HLE (with tools, text-only)', column 'Hy3'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; highest reasoning setting",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-11-05",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/tencent/Hy3",
          "name": "Hy3 card",
          "title": "tencent/Hy3 model card",
          "lab": "tencent",
          "date": "2026-07-06",
          "printed": "53.2",
          "where": "Model card, Appendix image, row HLE (with tools, text-only), column Hy3. Same table as the Hy3 page's appendix, but a different image file (the page's is d1af768b…). The card's top bar chart has the same bytes as the page's chart, so it is not cited. Mirrored with identical bytes on huggingface.co/tencent/Hy3-FP8 and github.com/Tencent-Hunyuan/Hy3 (assets/benchmark-appendix.png). Notes: highest reasoning tier; SWE-agent scaffold; Terminus-2, 4h timeout, 500 episodes; MCP-Atlas 500-task public set, 100-call budget, Gemini 2.5 Pro judge; BrowseComp self-summary context management."
        }
      ],
      "source_image": "https://frontierlog.ai/sources/d1af768b6734a711.png",
      "source_image_original": "https://hy-model-ap-prod-1258344703.cos.accelerate.myqcloud.com/llm-blog/default/9a5ea02f/2026070622113393_004bbe5190853d7c35825b9257672dde.png?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1783347098;2098707098&q-key-time=1783347098;2098707098&q-header-list=&q-url-param-list=&q-signature=507857321cef2b9fef5e4ef8a8f9ab8937d2480b"
    },
    {
      "id": "hy3:hle:with-tools-text-only:2",
      "model": "hy3",
      "model_name": "Hy3",
      "model_date": "2026-07-06",
      "event": "https://frontierlog.ai/event/2026-07-06-hy3",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 35.4,
      "unit": "%",
      "printed": "35.4",
      "reported_by": "tencent",
      "run_by": null,
      "label": "Hy3 preview",
      "source": "https://hy.tencent.ai/research/hy3",
      "source_title": "Introducing Hy3",
      "source_date": "2026-07-06",
      "where": "'Appendix: Benchmark Results' table image, row 'HLE (with tools, text-only)', column 'Hy3 preview'",
      "own": true,
      "primary": false,
      "parallel": false,
      "notes": "Hy3 preview, re-evaluated at Hy3 launch; text-only subset; highest reasoning setting",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-11-05",
      "corroborated_by": [
        {
          "url": "https://huggingface.co/tencent/Hy3",
          "name": "Hy3 card",
          "title": "tencent/Hy3 model card",
          "lab": "tencent",
          "date": "2026-07-06",
          "printed": "35.4",
          "where": "Model card, Appendix image, row HLE (with tools, text-only), column Hy3 preview. Same table as the Hy3 page's appendix, but a different image file (the page's is d1af768b…). The card's top bar chart has the same bytes as the page's chart, so it is not cited. Mirrored with identical bytes on huggingface.co/tencent/Hy3-FP8 and github.com/Tencent-Hunyuan/Hy3 (assets/benchmark-appendix.png). Notes: highest reasoning tier; SWE-agent scaffold; Terminus-2, 4h timeout, 500 episodes; MCP-Atlas 500-task public set, 100-call budget, Gemini 2.5 Pro judge; BrowseComp self-summary context management."
        }
      ],
      "source_image": "https://frontierlog.ai/sources/d1af768b6734a711.png",
      "source_image_original": "https://hy-model-ap-prod-1258344703.cos.accelerate.myqcloud.com/llm-blog/default/9a5ea02f/2026070622113393_004bbe5190853d7c35825b9257672dde.png?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1783347098;2098707098&q-key-time=1783347098;2098707098&q-header-list=&q-url-param-list=&q-signature=507857321cef2b9fef5e4ef8a8f9ab8937d2480b"
    },
    {
      "id": "hy3:hle:with-tools-text-only:3",
      "model": "hy3",
      "model_name": "Hy3",
      "model_date": "2026-07-06",
      "event": "https://frontierlog.ai/event/2026-07-06-hy3",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 51.9,
      "unit": "%",
      "printed": "51.9",
      "reported_by": "tencent",
      "run_by": null,
      "label": "Hy3",
      "source": "https://hy.tencent.ai/research/hy4-preview",
      "source_title": "Introducing Hy4 preview",
      "source_date": "2026-08-28",
      "where": "later report: Hy4 preview post, 'Appendix: Benchmark Results' table image, row 'HLE (with tools, text-only)', column 'Hy3'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "re-evaluated at Hy4 preview launch; text-only subset; highest reasoning setting",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-07",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/02c751bf51c52c52.jpg",
      "source_image_original": "https://hy-model-ap-prod-1258344703.cos.accelerate.myqcloud.com/llm-blog/default/9a5ea02f/2026082812342438_ecb98509e92e4be02a40ca358414133a.jpg?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1787891670;2103251670&q-key-time=1787891670;2103251670&q-header-list=&q-url-param-list=&q-signature=b6d0c3c4d6d7e0e5219d177ec730735185ccdd1f"
    },
    {
      "id": "hy4-preview:hle:with-tools-text-only:1",
      "model": "hy4-preview",
      "model_name": "Hy4 preview",
      "model_date": "2026-08-28",
      "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 57.3,
      "unit": "%",
      "printed": "57.3",
      "reported_by": "tencent",
      "run_by": null,
      "label": "Hy4 preview",
      "source": "https://hy.tencent.ai/research/hy4",
      "source_title": "Introducing Hy4",
      "source_date": "2026-11-04",
      "where": "later report: Hy4 post, 'Appendix: benchmark results' table image, row 'HLE-Full (with tools)', column 'Hy4 preview' (dagger: text-only subset)",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "re-evaluated at Hy4 launch; text-only subset",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-08",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/8fc24d0f588e6069.jpg",
      "source_image_original": "https://hy-model-ap-prod-1258344703.cos.ap-guangzhou.myqcloud.com/llm-blog/default/9a5ea02f/2026110402023072_85b416b6484a88ef96ef6edda563b560.jpg?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1793728954;2109088954&q-key-time=1793728954;2109088954&q-header-list=&q-url-param-list=&q-signature=747236201b67dd4e9872ed806c8de224d0ca57c1"
    },
    {
      "id": "hy4-preview:hle:with-tools-text-only:2",
      "model": "hy4-preview",
      "model_name": "Hy4 preview",
      "model_date": "2026-08-28",
      "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 55.4,
      "unit": "%",
      "printed": "55.4",
      "reported_by": "tencent",
      "run_by": null,
      "label": "Hy4 preview",
      "source": "https://hy.tencent.ai/research/hy4-preview",
      "source_title": "Introducing Hy4 preview",
      "source_date": "2026-08-28",
      "where": "'Appendix: Benchmark Results' table image, row 'HLE (with tools, text-only)', column 'Hy4 preview'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; highest reasoning setting",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-09",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_image": "https://frontierlog.ai/sources/02c751bf51c52c52.jpg",
      "source_image_original": "https://hy-model-ap-prod-1258344703.cos.accelerate.myqcloud.com/llm-blog/default/9a5ea02f/2026082812342438_ecb98509e92e4be02a40ca358414133a.jpg?q-sign-algorithm=sha1&q-ak=AKIDRl074nOsGdJ9zjMsCRWP3ShmgS3VtX4S&q-sign-time=1787891670;2103251670&q-key-time=1787891670;2103251670&q-header-list=&q-url-param-list=&q-signature=b6d0c3c4d6d7e0e5219d177ec730735185ccdd1f"
    },
    {
      "id": "nemotron-3-5-super-vl-120b-a12b:hle:with-tools-text-only:1",
      "model": "nemotron-3-5-super-vl-120b-a12b",
      "model_name": "Nemotron 3.5 Super VL 120B-A12B",
      "model_date": "2026-10-15",
      "event": "https://frontierlog.ai/event/2026-10-15-nemotron-3-5-super-vl-bolt",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 46.01,
      "unit": "%",
      "printed": "46.01",
      "reported_by": "nvidia",
      "run_by": null,
      "label": "N3.5-Super 120B-A12B",
      "source": "https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3.5-Super-and-Lightning-Technical-Report.pdf",
      "source_title": "Nemotron 3.5 Super and Lightning: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Models for Agentic AI",
      "source_date": "2026-10-15",
      "where": "Table 10 (PDF page 23), row 'HLE (with tools, text)', column 'N3.5-Super 120B-A12B' (first column)",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; Python and web search; separate internal harness",
      "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/ee443a24cf8af62e.png"
    },
    {
      "id": "longcat-2-5:hle:with-tools-text-only:1",
      "model": "longcat-2-5",
      "model_name": "LongCat-2.5",
      "model_date": "2026-11-01",
      "event": "https://frontierlog.ai/event/2026-11-01-longcat-2-5",
      "benchmark": "hle",
      "benchmark_name": "Humanity's Last Exam",
      "setting": "with-tools-text-only",
      "setting_name": "With tools, text-only subset",
      "score": 63.1,
      "unit": "%",
      "printed": "63.1",
      "reported_by": "meituan",
      "run_by": null,
      "label": "LongCat-2.5",
      "source": "https://longcat.ai/blog/longcat-2.5",
      "source_title": "LongCat-2.5: Native Multimodality for Agentic Tasks",
      "source_date": "2026-11-01",
      "where": "'Evaluations' section table at the end of the post (rendered by the page's script; load https://longcat.ai/blog/longcat-2.5 in a browser and scroll to the end), row 'HLE w/tools' under 'General Agent', column 'LongCat-2.5'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "text-only subset; Claude Code harness; 256K output budget",
      "caveat": "",
      "human_verified": true,
      "human_verified_on": "2026-11-02",
      "check": "human",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": [],
      "source_capture": "https://frontierlog.ai/captures/2b178d4c19a6dbe7.png"
    }
  ]
}