{
  "api_version": "1",
  "updated": "2026-10-28",
  "latest_event": "2026-10-27",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "gpt-oss-20b",
    "name": "gpt-oss-20b",
    "lab": "openai",
    "date": "2025-08-05",
    "source": "https://openai.com/index/introducing-gpt-oss",
    "source_title": "Introducing gpt-oss",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
    "releases": [],
    "variant_of": null,
    "made_by": "OpenAI",
    "url": "https://frontierlog.ai/model/gpt-oss-20b",
    "results": [
      {
        "id": "gpt-oss-20b:aime-2024:with-tools:1",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2024",
        "benchmark_name": "AIME 2024",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 96.0,
        "unit": "%",
        "printed": "96",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b",
        "source": "https://openai.com/index/introducing-gpt-oss",
        "source_title": "Introducing gpt-oss",
        "source_date": "2025-08-05",
        "where": "bar chart 'AIME 2024 (tools) / Competition math', bar 'gpt-oss-20b' (page screenshot (job 20879 capture), part 03)",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "high reasoning effort; tools",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2025:with-tools:1",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 98.7,
        "unit": "%",
        "printed": "98.7",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b",
        "source": "https://openai.com/index/introducing-gpt-oss",
        "source_title": "Introducing gpt-oss",
        "source_date": "2025-08-05",
        "where": "bar chart 'AIME 2025 (tools) / Competition math', bar 'gpt-oss-20b' (page screenshot (job 20879 capture), part 03)",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "high reasoning effort; tools",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2024:no-tools:1",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2024",
        "benchmark_name": "AIME 2024",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 42.1,
        "unit": "%",
        "printed": "42.1",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b low",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2024 (no tools)', column 'gpt-oss-20b' / 'low'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "low reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2024:no-tools:2",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2024",
        "benchmark_name": "AIME 2024",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 80.0,
        "unit": "%",
        "printed": "80.0",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b medium",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2024 (no tools)', column 'gpt-oss-20b' / 'medium'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "medium reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2024:no-tools:3",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2024",
        "benchmark_name": "AIME 2024",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 92.1,
        "unit": "%",
        "printed": "92.1",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b high",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2024 (no tools)', column 'gpt-oss-20b' / 'high'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "high reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2024:with-tools:2",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2024",
        "benchmark_name": "AIME 2024",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 61.2,
        "unit": "%",
        "printed": "61.2",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b low",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2024 (with tools)', column 'gpt-oss-20b' / 'low'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "low reasoning effort; tools",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2024:with-tools:3",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2024",
        "benchmark_name": "AIME 2024",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 86.0,
        "unit": "%",
        "printed": "86.0",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b medium",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2024 (with tools)', column 'gpt-oss-20b' / 'medium'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "medium reasoning effort; tools",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2025:no-tools:1",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 37.1,
        "unit": "%",
        "printed": "37.1",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b low",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2025 (no tools)', column 'gpt-oss-20b' / 'low'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "low reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2025:no-tools:2",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 72.1,
        "unit": "%",
        "printed": "72.1",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b medium",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2025 (no tools)', column 'gpt-oss-20b' / 'medium'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "medium reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2025:no-tools:3",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 91.7,
        "unit": "%",
        "printed": "91.7",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b high",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2025 (no tools)', column 'gpt-oss-20b' / 'high'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "high reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2025:with-tools:2",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 57.5,
        "unit": "%",
        "printed": "57.5",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b low",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2025 (with tools)', column 'gpt-oss-20b' / 'low'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "low reasoning effort; tools",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:aime-2025:with-tools:3",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "aime-2025",
        "benchmark_name": "AIME 2025",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 90.4,
        "unit": "%",
        "printed": "90.4",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b medium",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'AIME 2025 (with tools)', column 'gpt-oss-20b' / 'medium'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "medium reasoning effort; tools",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:swe-bench-verified:default:1",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "swe-bench-verified",
        "benchmark_name": "SWE-bench Verified",
        "setting": "default",
        "setting_name": "",
        "score": 37.4,
        "unit": "%",
        "printed": "37.4",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b low",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'SWE-Bench Verified', column 'gpt-oss-20b' / 'low'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "low reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:swe-bench-verified:default:2",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "swe-bench-verified",
        "benchmark_name": "SWE-bench Verified",
        "setting": "default",
        "setting_name": "",
        "score": 53.2,
        "unit": "%",
        "printed": "53.2",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b medium",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'SWE-Bench Verified', column 'gpt-oss-20b' / 'medium'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "medium reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:swe-bench-verified:default:3",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "swe-bench-verified",
        "benchmark_name": "SWE-bench Verified",
        "setting": "default",
        "setting_name": "",
        "score": 60.7,
        "unit": "%",
        "printed": "60.7",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss-20b high",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 2.6.1, Table 3 'Evaluations across multiple benchmarks and reasoning levels' (HTML), row 'SWE-Bench Verified', column 'gpt-oss-20b' / 'high'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "high reasoning effort",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "gpt-oss-20b:swe-bench-verified:default:4",
        "model": "gpt-oss-20b",
        "model_name": "gpt-oss-20b",
        "model_date": "2025-08-05",
        "event": "https://frontierlog.ai/event/2025-08-05-gpt-oss",
        "benchmark": "swe-bench-verified",
        "benchmark_name": "SWE-bench Verified",
        "setting": "default",
        "setting_name": "",
        "score": 60.0,
        "unit": "%",
        "printed": "60%",
        "reported_by": "openai",
        "run_by": null,
        "label": "gpt-oss 20b no browsing launch candidate",
        "source": "https://deploymentsafety.openai.com/gpt-oss",
        "source_title": "gpt-oss-120b & gpt-oss-20b Model Card",
        "source_date": "2025-08-05",
        "where": "section 5.2.3.1 SWE-bench Verified (N=477), SWE-bench Verified (n=477) bar chart image, bar 'gpt-oss 20b no browsing launch candidate'",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "preparedness run; no browsing; scaffold not stated; 477 of the 500 tasks",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "double_entry",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_image": "https://frontierlog.ai/sources/c9d7663e03e4d9f3.png",
        "source_image_original": "https://deploymentsafety.openai.com/data/eval-sets/gpt-oss/assets/gpt-oss_SWE-bench_Verified_(n=477).png"
      }
    ]
  }
}