{
  "api_version": "1",
  "updated": "2026-11-10",
  "latest_event": "2026-11-09",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "hy4-preview",
    "name": "Hy4 preview",
    "lab": "tencent",
    "date": "2026-08-28",
    "source": "https://hy.tencent.ai/research/hy4-preview",
    "source_title": "Introducing Hy4 preview",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
    "releases": [],
    "variant_of": null,
    "made_by": "Tencent",
    "url": "https://frontierlog.ai/model/hy4-preview",
    "results": [
      {
        "id": "hy4-preview:automationbench-1-0-6:unstated:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "automationbench-1-0-6",
        "benchmark_name": "AutomationBench 1.0.6",
        "setting": "unstated",
        "setting_name": "Set not stated",
        "score": 32.1,
        "unit": "%",
        "printed": "32.1",
        "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 'AutomationBench (v1.0.6)', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "AutomationBench v1.0.6",
        "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/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:terminal-bench-4-0:default:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "terminal-bench-4-0",
        "benchmark_name": "Terminal-Bench 4.0",
        "setting": "default",
        "setting_name": "",
        "score": 26.3,
        "unit": "%",
        "printed": "26.3",
        "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 'Terminal-Bench 4.0', column 'Hy4 preview'",
        "own": false,
        "primary": true,
        "parallel": false,
        "notes": "re-evaluated at Hy4 launch; Claude Code harness; up to 500 turns and a 13h timeout per trial",
        "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:automationbench-1-0-6:public:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "automationbench-1-0-6",
        "benchmark_name": "AutomationBench 1.0.6",
        "setting": "public",
        "setting_name": "Public set",
        "score": 36.3,
        "unit": "%",
        "printed": "36.3",
        "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 'AutomationBench v1.0.6', column 'Hy4 preview'",
        "own": false,
        "primary": true,
        "parallel": false,
        "notes": "re-evaluated at Hy4 launch; AutomationBench v1.0.6; official method",
        "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: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:terminal-bench-science-0-1:default:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "terminal-bench-science-0-1",
        "benchmark_name": "Terminal-Bench Science 0.1",
        "setting": "default",
        "setting_name": "",
        "score": 16.2,
        "unit": "%",
        "printed": "16.2",
        "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 'Terminal-Bench-Science 0.1', column 'Hy4 preview'",
        "own": false,
        "primary": true,
        "parallel": false,
        "notes": "re-evaluated at Hy4 launch; Claude Code harness; up to 500 turns and a 13h timeout per trial",
        "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:swe-bench-pro:unstated:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "swe-bench-pro",
        "benchmark_name": "SWE-bench Pro",
        "setting": "unstated",
        "setting_name": "Set not stated",
        "score": 65.7,
        "unit": "%",
        "printed": "65.7",
        "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 'SWE-bench Pro', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "SWE-agent scaffold; 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": "hy4-preview:deepswe-1-1:default:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "deepswe-1-1",
        "benchmark_name": "DeepSWE v1.1",
        "setting": "default",
        "setting_name": "",
        "score": 64.3,
        "unit": "%",
        "printed": "64.3",
        "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 'DeepSWE', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "mini-swe-agent harness; highest reasoning setting",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-11-09",
        "corroborated_by": [
          {
            "url": "https://huggingface.co/tencent/Hy4-preview",
            "name": "Hy4 preview card",
            "title": "tencent/Hy4-preview model card",
            "lab": "tencent",
            "date": "2026-08-28",
            "printed": "64.3",
            "where": "'A New Flagship Generation' bar chart (assets/benchmark.jpg), panel 'DeepSWE', Hy4 preview (dark blue) bar"
          },
          {
            "url": "https://hy.tencent.ai/research/hy4",
            "name": "Hy4",
            "title": "Introducing Hy4",
            "lab": "tencent",
            "date": "2026-11-04",
            "printed": "64.3",
            "where": "Appendix: benchmark results table (大Bench图表.jpg), row 'DeepSWE v1.1', column 'Hy4 preview'"
          }
        ],
        "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:terminal-bench-2-1:default:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "terminal-bench-2-1",
        "benchmark_name": "Terminal-Bench 2.1",
        "setting": "default",
        "setting_name": "",
        "score": 85.4,
        "unit": "%",
        "printed": "85.4",
        "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 'Terminal-Bench 2.1', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "Claude Code harness; up to 500 turns and a 12h timeout per trial; highest reasoning setting",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-11-09",
        "corroborated_by": [
          {
            "url": "https://huggingface.co/tencent/Hy4-preview",
            "name": "Hy4 preview card",
            "title": "tencent/Hy4-preview model card",
            "lab": "tencent",
            "date": "2026-08-28",
            "printed": "85.4",
            "where": "'A New Flagship Generation' bar chart (assets/benchmark.jpg), panel 'Terminal Bench 2.1', Hy4 preview (dark blue) bar"
          }
        ],
        "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:cybergym:default:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "cybergym",
        "benchmark_name": "CyberGym",
        "setting": "default",
        "setting_name": "",
        "score": 78.4,
        "unit": "%",
        "printed": "78.4",
        "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 'CyberGym', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "harness not stated; 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": "hy4-preview:mcp-atlas:public:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "mcp-atlas",
        "benchmark_name": "MCP Atlas",
        "setting": "public",
        "setting_name": "Public set (500 tasks)",
        "score": 83.7,
        "unit": "%",
        "printed": "83.7",
        "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 'MCP-Atlas (public)', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "500-task public set; 100-tool-call budget; Gemini 3.1 Pro Preview judge; 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": "hy4-preview:agents-last-exam:cli-pass:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "agents-last-exam",
        "benchmark_name": "Agents' Last Exam v1",
        "setting": "cli-pass",
        "setting_name": "CLI subset, pass rate",
        "score": 22.8,
        "unit": "%",
        "printed": "22.8",
        "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 'Agents' Last Exam (ALE-CLI)', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "105-task CLI subset; Claude Code harness; up to 12h per task; highest reasoning setting",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-11-09",
        "corroborated_by": [
          {
            "url": "https://huggingface.co/tencent/Hy4-preview",
            "name": "Hy4 preview card",
            "title": "tencent/Hy4-preview model card",
            "lab": "tencent",
            "date": "2026-08-28",
            "printed": "22.8",
            "where": "'A New Flagship Generation' bar chart (assets/benchmark.jpg), panel 'Agents' Last Exam (ALE-CLI)', Hy4 preview (dark blue) bar"
          }
        ],
        "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: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": "hy4-preview:gpqa-diamond:default:1",
        "model": "hy4-preview",
        "model_name": "Hy4 preview",
        "model_date": "2026-08-28",
        "event": "https://frontierlog.ai/event/2026-08-28-hy4-preview",
        "benchmark": "gpqa-diamond",
        "benchmark_name": "GPQA Diamond",
        "setting": "default",
        "setting_name": "",
        "score": 92.3,
        "unit": "%",
        "printed": "92.3",
        "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 'GPQA Diamond', column 'Hy4 preview'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "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": "hy4-preview:hle:no-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": "no-tools-text-only",
        "setting_name": "No tools, text-only subset",
        "score": 43.4,
        "unit": "%",
        "printed": "43.4",
        "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 (no tools, text-only)', column 'Hy4 preview'",
        "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": "2026-11-09",
        "corroborated_by": [
          {
            "url": "https://huggingface.co/tencent/Hy4-preview",
            "name": "Hy4 preview card",
            "title": "tencent/Hy4-preview model card",
            "lab": "tencent",
            "date": "2026-08-28",
            "printed": "43.4",
            "where": "'A New Flagship Generation' bar chart (assets/benchmark.jpg), panel 'Humanity's Last Exam (text-only, no tools)', Hy4 preview (dark blue) bar"
          }
        ],
        "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"
      }
    ],
    "price": [
      {
        "from": "2026-08-28",
        "listed": false,
        "currency": "USD",
        "input_per_1m": 0.834,
        "cached_input_per_1m": 0.042,
        "cached_input_derived": false,
        "output_per_1m": 2.501,
        "cache": {
          "kind": "price"
        },
        "tiers": [],
        "printed": {
          "input": "$0.834",
          "cached_input": "$0.042",
          "output": "$2.501"
        },
        "printed_unit": "1M tokens",
        "notes": "",
        "sources": [
          "https://hy.tencent.ai/research/hy4-preview",
          "https://openrouter.ai/tencent/hy4-preview",
          "https://www.tencentcloud.com/document/product/1300/78937"
        ],
        "check": "corroborated"
      }
    ],
    "context": {
      "from": "2026-08-28",
      "listed": false,
      "window_tokens": 1000000,
      "input_tokens": null,
      "output_tokens": null,
      "tiers": [],
      "printed": {
        "window": "1M",
        "input": null,
        "output": null
      },
      "notes": "",
      "sources": [
        "https://hy.tencent.ai/research/hy4-preview"
      ],
      "check": "human"
    },
    "prices": [
      {
        "from": "2026-08-28",
        "listed": false,
        "currency": "USD",
        "input_per_1m": 0.834,
        "cached_input_per_1m": 0.042,
        "cached_input_derived": false,
        "output_per_1m": 2.501,
        "cache": {
          "kind": "price"
        },
        "tiers": [],
        "printed": {
          "input": "$0.834",
          "cached_input": "$0.042",
          "output": "$2.501"
        },
        "printed_unit": "1M tokens",
        "notes": "",
        "sources": [
          "https://hy.tencent.ai/research/hy4-preview",
          "https://openrouter.ai/tencent/hy4-preview",
          "https://www.tencentcloud.com/document/product/1300/78937"
        ],
        "check": "corroborated"
      }
    ],
    "context_history": [
      {
        "from": "2026-08-28",
        "listed": false,
        "window_tokens": 1000000,
        "input_tokens": null,
        "output_tokens": null,
        "tiers": [],
        "printed": {
          "window": "1M",
          "input": null,
          "output": null
        },
        "notes": "",
        "sources": [
          "https://hy.tencent.ai/research/hy4-preview"
        ],
        "check": "human"
      }
    ],
    "stories": [],
    "compare": []
  }
}