{
  "api_version": "1",
  "updated": "2026-10-26",
  "latest_event": "2026-10-24",
  "rights": "© Frontier Log, all rights reserved.",
  "model": {
    "slug": "claude-haiku-5-5",
    "name": "Claude Haiku 5.5",
    "lab": "anthropic",
    "date": "2026-10-07",
    "source": "https://www.anthropic.com/claude-haiku-5-5",
    "source_title": "Introducing Claude Haiku 5.5",
    "flagship": false,
    "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
    "releases": [],
    "variant_of": null,
    "made_by": "Anthropic",
    "url": "https://frontierlog.ai/model/claude-haiku-5-5",
    "results": [
      {
        "id": "claude-haiku-5-5:hle:no-tools:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "no-tools",
        "setting_name": "No tools",
        "score": 45.9,
        "unit": "%",
        "printed": "45.9%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5",
        "source_title": "Introducing Claude Haiku 5.5",
        "source_date": "2026-10-07",
        "where": "Performance table, row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Haiku 5.5', 'no tools'",
        "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-haiku-5-5-system-card",
            "name": "Claude Haiku 5.5 card",
            "title": "Claude Haiku 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-10-07",
            "printed": "45.9",
            "where": "Table 8.1.A (p.111), row Humanity's Last Exam / No tools, column Haiku 5.5"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/119b9e1f2a7a6ddd.png"
      },
      {
        "id": "claude-haiku-5-5:swe-bench-pro:default:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "swe-bench-pro",
        "benchmark_name": "SWE-bench Pro",
        "setting": "default",
        "setting_name": "",
        "score": 64.8,
        "unit": "%",
        "printed": "64.8",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5-system-card",
        "source_title": "Claude Haiku 5.5 System Card",
        "source_date": "2026-10-07",
        "where": "Section 8.1 Evaluation summary, Table 8.1.A, row 'SWE-Bench Pro', column 'Haiku 5.5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "adaptive thinking, max effort; average of 5 trials",
        "caveat": "The source doesn't say whether this is the public or the commercial set",
        "human_verified": true,
        "human_verified_on": "2026-10-17",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": [],
        "source_capture": "https://frontierlog.ai/captures/1189b487a0e11054.png"
      },
      {
        "id": "claude-haiku-5-5:terminal-bench-4-0:default:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "terminal-bench-4-0",
        "benchmark_name": "Terminal-Bench 4.0",
        "setting": "default",
        "setting_name": "",
        "score": 39.2,
        "unit": "%",
        "printed": "39.2%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5",
        "source_title": "Introducing Claude Haiku 5.5",
        "source_date": "2026-10-07",
        "where": "benchmark table, row 'Terminal-Bench 4.0', column 'Haiku 5.5'",
        "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-haiku-5-5-system-card",
            "name": "Claude Haiku 5.5 card",
            "title": "Claude Haiku 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-10-07",
            "printed": "39.2%",
            "where": "Section 8.4 (p.115-116) sentence: \"Claude Haiku 5.5 scored 39.2% on Terminal-Bench 4.0 with safeguards enabled\" (no fallback model)"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/119b9e1f2a7a6ddd.png"
      },
      {
        "id": "claude-haiku-5-5:hle:with-tools:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "hle",
        "benchmark_name": "Humanity's Last Exam",
        "setting": "with-tools",
        "setting_name": "With tools",
        "score": 57.4,
        "unit": "%",
        "printed": "57.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5",
        "source_title": "Introducing Claude Haiku 5.5",
        "source_date": "2026-10-07",
        "where": "Performance table (HTML), row 'Multidisciplinary reasoning, Humanity's Last Exam', column 'Haiku 5.5', the 'with tools' value",
        "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://www.anthropic.com/claude-haiku-5-5-system-card",
            "name": "Claude Haiku 5.5 card",
            "title": "Claude Haiku 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-10-07",
            "printed": "57.4",
            "where": "Table 8.1.A, Humanity's Last Exam With tools row, Haiku 5.5 column"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/119b9e1f2a7a6ddd.png"
      },
      {
        "id": "claude-haiku-5-5:terminal-bench-science-0-1:default:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "terminal-bench-science-0-1",
        "benchmark_name": "Terminal-Bench Science 0.1",
        "setting": "default",
        "setting_name": "",
        "score": 20.6,
        "unit": "%",
        "printed": "20.6%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5-system-card",
        "source_title": "Claude Haiku 5.5 System Card",
        "source_date": "2026-10-07",
        "where": "Section 8.5 'Terminal-Bench-Science 0.1', first paragraph after the task description",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "Claude Code in --bare mode; max effort; no fallback model, so safeguard stops scored as failures; standard error ±3.7 points",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-19",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "claude-haiku-5-5:frontiercode-1-1:main:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "frontiercode-1-1",
        "benchmark_name": "FrontierCode 1.1",
        "setting": "main",
        "setting_name": "Main set",
        "score": 46.4,
        "unit": "%",
        "printed": "46.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5",
        "source_title": "Introducing Claude Haiku 5.5",
        "source_date": "2026-10-07",
        "where": "Performance table (HTML), row 'Agentic coding, FrontierCode 1.1 (Main)', column 'Haiku 5.5'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort; run by Cognition in Claude Code; mean of 5 runs",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-19",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-haiku-5-5-system-card",
            "name": "Claude Haiku 5.5 card",
            "title": "Claude Haiku 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-10-07",
            "printed": "46.4",
            "where": "Table 8.1.A, row FrontierCode 1.1 (Main), column Haiku 5.5, '(max)' (also 8.3 text: '46.4% on Main')"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/119b9e1f2a7a6ddd.png"
      },
      {
        "id": "claude-haiku-5-5:frontiercode-1-1:main:2",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "frontiercode-1-1",
        "benchmark_name": "FrontierCode 1.1",
        "setting": "main",
        "setting_name": "Main set",
        "score": 45.8,
        "unit": "%",
        "printed": "45.8",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5 (xhigh)",
        "source": "https://www.anthropic.com/claude-haiku-5-5-system-card",
        "source_title": "Claude Haiku 5.5 System Card",
        "source_date": "2026-10-07",
        "where": "Section 8.1, Table 8.1.A, row 'FrontierCode 1.1 (Main)', column 'Haiku 5.5', the '(xhigh)' value",
        "own": true,
        "primary": false,
        "parallel": false,
        "notes": "xhigh effort; run by Cognition in Claude Code; mean of 5 runs",
        "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/1189b487a0e11054.png"
      },
      {
        "id": "claude-haiku-5-5:frontiercode-1-1:extended:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "frontiercode-1-1",
        "benchmark_name": "FrontierCode 1.1",
        "setting": "extended",
        "setting_name": "Extended set",
        "score": 58.4,
        "unit": "%",
        "printed": "58.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5-system-card",
        "source_title": "Claude Haiku 5.5 System Card",
        "source_date": "2026-10-07",
        "where": "Section 8.3 'FrontierCode', paragraph 'Claude Haiku 5.5 scored best at max effort ...', Extended figure",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "max effort; run by Cognition in Claude Code; mean of 5 runs",
        "caveat": "",
        "human_verified": true,
        "human_verified_on": "2026-10-20",
        "check": "human",
        "corroborated": false,
        "corroborated_on": null,
        "corroborated_by": []
      },
      {
        "id": "claude-haiku-5-5:osworld-2-1:offline-partial:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "osworld-2-1",
        "benchmark_name": "OSWorld 2.1",
        "setting": "offline-partial",
        "setting_name": "Offline set, partial score",
        "score": 72.4,
        "unit": "%",
        "printed": "72.4%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5",
        "source_title": "Introducing Claude Haiku 5.5",
        "source_date": "2026-10-07",
        "where": "benchmark table (HTML), row 'Computer use / OSWorld 2.1', column 'Haiku 5.5', the value tagged 'Offline subset'",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "82-task offline subset, no internet access; max effort; mean of 5 attempts",
        "caveat": "",
        "human_verified": false,
        "human_verified_on": null,
        "check": "corroborated",
        "corroborated": true,
        "corroborated_on": "2026-10-20",
        "corroborated_by": [
          {
            "url": "https://www.anthropic.com/claude-haiku-5-5-system-card",
            "name": "Claude Haiku 5.5 card",
            "title": "Claude Haiku 5.5 System Card",
            "lab": "anthropic",
            "date": "2026-10-07",
            "printed": "72.4%",
            "where": "section 8.9.3 OSWorld 2.1 offline subset (p. 128), sentence: 'Claude Haiku 5.5 achieved a partial score of 72.4% and a strict pass rate of 37.1%'; also 8.1 Evaluation summary, row 'OSWorld 2.1 (offline subset)', column Haiku 5.5: '72.4'"
          }
        ],
        "source_capture": "https://frontierlog.ai/captures/119b9e1f2a7a6ddd.png"
      },
      {
        "id": "claude-haiku-5-5:osworld-2-1:offline-strict:1",
        "model": "claude-haiku-5-5",
        "model_name": "Claude Haiku 5.5",
        "model_date": "2026-10-07",
        "event": "https://frontierlog.ai/event/2026-10-07-claude-haiku-5-5",
        "benchmark": "osworld-2-1",
        "benchmark_name": "OSWorld 2.1",
        "setting": "offline-strict",
        "setting_name": "Offline set, strict pass rate",
        "score": 37.1,
        "unit": "%",
        "printed": "37.1%",
        "reported_by": "anthropic",
        "run_by": null,
        "label": "Claude Haiku 5.5",
        "source": "https://www.anthropic.com/claude-haiku-5-5-system-card",
        "source_title": "Claude Haiku 5.5 System Card",
        "source_date": "2026-10-07",
        "where": "section 8.9.3 OSWorld 2.1 offline subset, text (PDF page 128), the Claude Haiku 5.5 strict pass rate",
        "own": true,
        "primary": true,
        "parallel": false,
        "notes": "82-task offline subset, no internet access; max effort; mean of 5 attempts",
        "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/78b2fca2e8e996c4.png"
      }
    ]
  }
}