gpt.college

Anthropic · Claude Fable · Reviewed 2026-09-02

Claude Fable 5.1

Anthropic's latest Fable model for long-running coding and knowledge work, with adaptive always-on thinking.

All models
Overall · 0–10086.3
Status
active
Released
2026-09-01
Context
1,000,000 tokens
Parameters
Undisclosed
Inputs
Text · Image
Type
multimodal
Max output
128,000 tokens
API model ID
claude-fable-5-1
Reasoning
adaptive
Knowledge cutoff
2026-06
Training cutoff
2026-06
Weights / license
Unavailable
Price / 1M tokens
$10 in · $50 out
Cached input
$0.25
1M-token context128K maximum outputStrong coding and professional-work benchmark coverage
CodingReasoningResearchMultimodalAgentic

Independent benchmark scores

Raw benchmark results

Source-backed rows with the original evaluation context preserved.

21 results
BenchmarkScoreSourceEvaluation notes
AutomationBench-AAChecked 2026-10-02 59% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AA GDP.pdfChecked 2026-10-02 26% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
GDPval-AA v2.1Checked 2026-10-02 1735 Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AA Humanity's Last ExamChecked 2026-10-02 59% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
Artificial Analysis Intelligence Index v4.3.2Checked 2026-10-02 53 Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AA Terminal-Bench 4.0Checked 2026-10-02 52% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
Artificial Analysis Intelligence IndexChecked 2026-09-05 66 Artificial Analysis Artificial Analysis Intelligence Index v4.x; reasoning=max.
Artificial Analysis Intelligence IndexEvaluated 2026-09-02 66 Artificial Analysis Artificial Analysis Intelligence Index v4.1.1; Claude Fable 5.1 max effort with fallback.
Artificial Analysis Intelligence IndexEvaluated 2026-09-02 65 Artificial Analysis Artificial Analysis Intelligence Index v4.1.1; Claude Fable 5.1 xhigh effort with fallback.
Artificial Analysis Intelligence IndexEvaluated 2026-09-02 62 Artificial Analysis Artificial Analysis Intelligence Index v4.1.1; Claude Fable 5.1 high effort with fallback.
Artificial Analysis Intelligence IndexEvaluated 2026-09-02 60 Artificial Analysis Artificial Analysis Intelligence Index v4.1.1; Claude Fable 5.1 medium effort with fallback.
Artificial Analysis Intelligence IndexEvaluated 2026-09-02 58 Artificial Analysis Artificial Analysis Intelligence Index v4.1.1; Claude Fable 5.1 low effort with fallback.
AutomationBenchEvaluated 2026-09-01 31.4% Anthropic Anthropic-reported AutomationBench result; version and harness are unspecified in the supplied benchmark table; production safeguards enabled.
CursorBench 3.2.0Evaluated 2026-09-01 73.4% Anthropic Anthropic-reported; benchmark version=CursorBench 3.2.0.
GDPval-AA v2Evaluated 2026-09-01 1853 Anthropic Anthropic-reported GDPval-AA v2 score; score unit is Elo.
Humanity's Last Exam (HLE)Evaluated 2026-09-01 60.9% Anthropic Anthropic-reported Humanity's Last Exam result; tools=none; production safeguards enabled.
Humanity's Last Exam with ToolsEvaluated 2026-09-01 65% Anthropic Anthropic-reported Humanity's Last Exam result; tools=enabled; production safeguards enabled.
OSWorld 2.0 PartialEvaluated 2026-09-01 77.9% Anthropic Anthropic-reported OSWorld 2.0 Partial result; uses the benchmark authors' August 2026 task release and is not directly comparable with earlier task files.
OSWorld 2.0 StrictEvaluated 2026-09-01 41.7% Anthropic Anthropic-reported OSWorld 2.0 Strict result; uses the benchmark authors' August 2026 task release and is not directly comparable with earlier task files.
Terminal-Bench 4.0Evaluated 2026-09-01 55.8% Anthropic Anthropic-reported; benchmark version=Terminal-Bench 4.0; harness details remain source-specific.
Terminal-Bench-Science 0.1Evaluated 2026-09-01 52.6% Anthropic Anthropic-reported; benchmark version=Terminal-Bench-Science 0.1; standard error is approximately ±3.5–4.5 points per model; Claude Code harness.

Strengths

  • 1M-token context
  • 128K maximum output
  • Strong coding and professional-work benchmark coverage

Limitations

  • Parameter count and weights are not publicly disclosed
  • Production safeguards and tool harness affect reported scores

Model evidence · official · Checked 2026-09-02

Claude Fable 5.1

Official Anthropic Fable page for model availability, API ID, pricing, context/output limits, thinking behavior, and benchmark table notes.

Task lenses preserve raw scores. Radar axes use score / current leader × 100; overall is the equal-weight average of all five indices. Missing evidence is not a zero or a low rank.