{
  "api_version": "1",
  "updated": "2026-10-30",
  "latest_event": "2026-10-29",
  "rights": "© Frontier Log, all rights reserved.",
  "benchmark": {
    "slug": "sec-bench-pro-no-release",
    "name": "SEC-bench Pro, no release named",
    "family": "sec-bench-pro",
    "unit": "%",
    "higher_is_better": true,
    "status": "active",
    "active": true,
    "superseded_by": null,
    "retired": null,
    "benchmark_url": "https://sec-bench.github.io",
    "introduced": "2026-05-01",
    "creator": "University of Illinois Urbana-Champaign",
    "introduced_source": "https://sec-bench.github.io",
    "settings": [
      {
        "setting": "default",
        "name": "",
        "url": "https://frontierlog.ai/benchmark/sec-bench-pro-no-release",
        "results": 7
      }
    ]
  },
  "total": 7,
  "results": [
    {
      "id": "gpt-5-5:sec-bench-pro-no-release:default: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": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 45.8,
      "unit": "%",
      "printed": "45.8%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.5",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "cyber eval table (HTML; columns GPT‑5.6 Sol, GPT‑5.6 Sol Ultra, GPT‑5.6 Terra, GPT‑5.6 Luna, GPT‑5.5, ...; rows Capture-the-Flag Challenges, SEC-Bench Pro, ExploitBench, ExploitGym), row 'SEC-Bench Pro', column 'GPT‑5.5'",
      "own": false,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-5-6-luna:sec-bench-pro-no-release:default:1",
      "model": "gpt-5-6-luna",
      "model_name": "GPT-5.6 Luna",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 48.9,
      "unit": "%",
      "printed": "48.9%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Luna",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "cyber eval table (HTML; columns GPT‑5.6 Sol, GPT‑5.6 Sol Ultra, GPT‑5.6 Terra, GPT‑5.6 Luna, GPT‑5.5, ...; rows Capture-the-Flag Challenges, SEC-Bench Pro, ExploitBench, ExploitGym), row 'SEC-Bench Pro', column 'GPT‑5.6 Luna'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-5-6-sol:sec-bench-pro-no-release:default:1",
      "model": "gpt-5-6-sol",
      "model_name": "GPT-5.6 Sol",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 71.2,
      "unit": "%",
      "printed": "71.2%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Sol",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "cyber eval table (HTML; columns GPT‑5.6 Sol, GPT‑5.6 Sol Ultra, GPT‑5.6 Terra, GPT‑5.6 Luna, GPT‑5.5, ...; rows Capture-the-Flag Challenges, SEC-Bench Pro, ExploitBench, ExploitGym), row 'SEC-Bench Pro', column 'GPT‑5.6 Sol'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-5-6-sol:sec-bench-pro-no-release:default:2",
      "model": "gpt-5-6-sol",
      "model_name": "GPT-5.6 Sol",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 74.3,
      "unit": "%",
      "printed": "74.3%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Sol Ultra",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "cyber eval table (HTML; columns GPT‑5.6 Sol, GPT‑5.6 Sol Ultra, GPT‑5.6 Terra, GPT‑5.6 Luna, GPT‑5.5, ...; rows Capture-the-Flag Challenges, SEC-Bench Pro, ExploitBench, ExploitGym), row 'SEC-Bench Pro', column 'GPT‑5.6 Sol Ultra'",
      "own": true,
      "primary": false,
      "parallel": true,
      "notes": "ultra mode: four parallel agents",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-5-6-sol:sec-bench-pro-no-release:default:3",
      "model": "gpt-5-6-sol",
      "model_name": "GPT-5.6 Sol",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 79.1,
      "unit": "%",
      "printed": "79.1%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Sol",
      "source": "https://openai.com/index/gpt-6-astra",
      "source_title": "GPT-6 Astra: A new generation of intelligence",
      "source_date": "2026-09-03",
      "where": "Cybersecurity eval table (HTML; columns GPT‑6 Astra, GPT‑5.6 Sol, Claude Fable 5.1, ...), row 'SEC-Bench Pro', column 'GPT‑5.6 Sol'",
      "own": false,
      "primary": false,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-5-6-terra:sec-bench-pro-no-release:default:1",
      "model": "gpt-5-6-terra",
      "model_name": "GPT-5.6 Terra",
      "model_date": "2026-07-09",
      "event": "https://frontierlog.ai/event/2026-07-09-gpt-5-6",
      "benchmark": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 57.7,
      "unit": "%",
      "printed": "57.7%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑5.6 Terra",
      "source": "https://openai.com/index/gpt-5-6",
      "source_title": "GPT-5.6: Frontier intelligence that scales with your ambition",
      "source_date": "2026-07-09",
      "where": "cyber eval table (HTML; columns GPT‑5.6 Sol, GPT‑5.6 Sol Ultra, GPT‑5.6 Terra, GPT‑5.6 Luna, GPT‑5.5, ...; rows Capture-the-Flag Challenges, SEC-Bench Pro, ExploitBench, ExploitGym), row 'SEC-Bench Pro', column 'GPT‑5.6 Terra'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "double_entry",
      "corroborated": false,
      "corroborated_on": null,
      "corroborated_by": []
    },
    {
      "id": "gpt-6-astra:sec-bench-pro-no-release:default:1",
      "model": "gpt-6-astra",
      "model_name": "GPT-6 Astra",
      "model_date": "2026-09-03",
      "event": "https://frontierlog.ai/event/2026-09-03-gpt-6-astra",
      "benchmark": "sec-bench-pro-no-release",
      "benchmark_name": "SEC-bench Pro, no release named",
      "setting": "default",
      "setting_name": "",
      "score": 85.4,
      "unit": "%",
      "printed": "85.4%",
      "reported_by": "openai",
      "run_by": null,
      "label": "GPT‑6 Astra",
      "source": "https://openai.com/index/gpt-6-astra",
      "source_title": "GPT-6 Astra: A new generation of intelligence",
      "source_date": "2026-09-03",
      "where": "Cybersecurity eval table (HTML; columns GPT‑6 Astra, GPT‑5.6 Sol, Claude Fable 5.1, ...), row 'SEC-Bench Pro', column 'GPT‑6 Astra'",
      "own": true,
      "primary": true,
      "parallel": false,
      "notes": "",
      "caveat": "",
      "human_verified": false,
      "human_verified_on": null,
      "check": "corroborated",
      "corroborated": true,
      "corroborated_on": "2026-10-30",
      "corroborated_by": [
        {
          "url": "https://storage.googleapis.com/deepmind-media/gemini/gemini_4_carbon_model_evaluation.pdf",
          "name": "Gemini 4 Carbon card",
          "title": "Gemini 4 Carbon Model evaluation: approach, methodology & results",
          "lab": "google",
          "date": "2026-10-28",
          "printed": "85.4%",
          "where": "page 5 'Results' table (raster image), row 'SEC-bench Pro', column 'GPT-6 Astra' (Google restating; footnote: competitor numbers taken from their published system cards and technical reports)"
        },
        {
          "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-carbon/",
          "name": "Gemini 4 Carbon",
          "title": "Gemini 4 Carbon: our next frontier model",
          "lab": "google",
          "date": "2026-10-28",
          "printed": "85.4%",
          "where": "SEC-bench Pro bar chart (Pass@1), value label on bar 'GPT-6 Astra'"
        }
      ]
    }
  ]
}