Claude Fable 5 vs Claude Fable 5.1
Different efforts: max · medium · Compare all at high
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Fable 555.5
Fable 5.153.8
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,504
Fable 5.1 · max1,502
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,542.3
Fable 5.1 · max1,585.6
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,499
Fable 5.1 · max1,504.9
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,503.6
Fable 5.1 · max1,527.3
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,486.7
Fable 5.1 · max1,501.7
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,502.3
Fable 5.1 · max1,517.2
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,506.4
Fable 5.1 · max1,560.1
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,491
Fable 5.1 · max1,520.8
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,480.7
Fable 5.1 · max1,507.5
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,493.4
Fable 5.1 · max1,524.3
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,503
Fable 5.1 · max1,518.6
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
Fable 5 · high1,495.8
Fable 5.1 · max1,520.5
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
Fable 543.3
Fable 5.137.6
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
Fable 5 · xhigh70.7
Fable 5.1 · max70.8
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
Fable 528.6
Fable 5.129.1
AA Intelligence Index 4.3.2 AA Intelligence Index 4.3.2: Artificial Analysis combines tests of knowledge, reasoning, coding and tool use into one score
Fable 549.6
Fable 5.148.9
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
Fable 565.4
Fable 5.163.1
AA Omniscience · Hallucination Rate AA Omniscience · Hallucination Rate: Measures how often attempted answers are incorrect on the AA Omniscience factual knowledge test · lower is better
Fable 563.6
Fable 5.169.1
AA-LCR (long context) 1.1 AA-LCR (long context) 1.1: Questions that need reading and connecting facts across very long documents (around 100K tokens)
Fable 582.3
Fable 5.184.7
GDP.pdf · AA all-pass GDP.pdf · AA all-pass: Models reason over professional PDF documents; an attempt passes only when every atomic criterion passes
Fable 524
Fable 5.126.8
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete
Fable 564.4
Fable 5.1 · max71.1
ARC-AGI-1 · public evaluation ARC-AGI-1 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Fable 597.6
Fable 5.196.6
ARC-AGI-1 · semi-private ARC-AGI-1 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Fable 598.5
Fable 5.194.5
ARC-AGI-2 · public evaluation ARC-AGI-2 · public-eval: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Fable 596.4
Fable 5.187.5
ARC-AGI-2 · semi-private ARC-AGI-2 · semi-private: Models infer transformations from examples and solve unseen grid puzzles; the evaluation split is explicit
Fable 589.2
Fable 5.186.3
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
Fable 547
Fable 5.1 · max48
FrontierMath Erdos FrontierMath Erdos: Models solve expert-written advanced mathematics problems on the named tier and edition
Fable 50
Fable 5.1 · max0
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Fable 590.2
Fable 5.1 · max87.8
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Fable 587
Fable 5.1 · max90.2
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
Fable 552
Fable 5.1 · max58
MathArena · ArXivMath · April 2026 MathArena · arxiv · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 572.5
Fable 5.1 · max77.5
MathArena · ArXivMath · June 2026 MathArena · arxiv · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 585.4
Fable 5.1 · max91
MathArena · ArXivMath · May 2026 MathArena · arxiv · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 587.5
Fable 5.1 · max87.5
MathArena · BrokenArXiv · June 2026 MathArena · arxiv_false · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 547.8
Fable 5.1 · low86.6
MathArena · BrokenArXiv · May 2026 MathArena · arxiv_false · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 544.5
Fable 5.1 · low78
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
Fable 588.6
Fable 5.188.5
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
Fable 556.1
Fable 5.153.5
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
Fable 588.5
Fable 5.191.3
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass
Fable 549.5
Fable 5.155.3
IFBench (instructions) IFBench (instructions): Checks whether the model follows precise formatting and content instructions
Fable 563.5
Fable 5.1—
EQ-Bench 3 EQ-Bench 3: Judge-rated responses to social and emotionally nuanced scenarios, scored through the source pairwise ranking
Fable 52,049.7
Fable 5.1—
MathArena · ArXivMath · August 2026 MathArena · arxiv · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5—
Fable 5.1 · high87.7
MathArena · BrokenArXiv · April 2026 MathArena · arxiv_false · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 554.1
Fable 5.1—
MathArena · BrokenArXiv · August 2026 MathArena · arxiv_false · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5—
Fable 5.1 · high79.8
MathArena · ArXivLean · June 2026 MathArena · arxivlean · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 515.2
Fable 5.1—
| Benchmark | Fable 5 | Fable 5.1 |
|---|---|---|
| Humanity's Last Exam | 55.5 | 53.8 |
| Arena Text · overall | 1,504 · high | 1,502 · max |
| Arena Text · Chinese | 1,542.3 · high | 1,585.6 · max |
| Arena Text · English | 1,499 · high | 1,504.9 · max |
| Arena Text · French | 1,503.6 · high | 1,527.3 · max |
| Arena Text · German | 1,486.7 · high | 1,501.7 · max |
| Arena Text · instruction following | 1,502.3 · high | 1,517.2 · max |
| Arena Text · Japanese | 1,506.4 · high | 1,560.1 · max |
| Arena Text · Korean | 1,491 · high | 1,520.8 · max |
| Arena Text · non-English | 1,480.7 · high | 1,507.5 · max |
| Arena Text · Polish | 1,493.4 · high | 1,524.3 · max |
| Arena Text · Russian | 1,503 · high | 1,518.6 · max |
| Arena Text · Spanish | 1,495.8 · high | 1,520.5 · max |
| AA Omniscience Index | 43.3 | 37.6 |
| SimpleQA Verified v1 | 70.7 · xhigh | 70.8 · max |
| CritPt (physics) | 28.6 | 29.1 |
| AA Intelligence Index 4.3.2 | 49.6 | 48.9 |
| AA Omniscience · Accuracy | 65.4 | 63.1 |
| AA Omniscience · Hallucination Rate | 63.6 | 69.1 |
| AA-LCR (long context) 1.1 | 82.3 | 84.7 |
| GDP.pdf · AA all-pass | 24 | 26.8 |
| MLCR-AA | 64.4 | 71.1 · max |
| ARC-AGI-1 · public evaluation | 97.6 | 96.6 |
| ARC-AGI-1 · semi-private | 98.5 | 94.5 |
| ARC-AGI-2 · public evaluation | 96.4 | 87.5 |
| ARC-AGI-2 · semi-private | 89.2 | 86.3 |
| Chess Puzzles | 47 | 48 · max |
| FrontierMath Erdos | 0 | 0 · max |
| FrontierMath Tier 4 (v2) | 90.2 | 87.8 · max |
| FrontierMath Tiers 1–3 (v2) | 87 | 90.2 · max |
| Mystery Game Puzzles | 52 | 58 · max |
| MathArena · ArXivMath · April 2026 | 72.5 | 77.5 · max |
| MathArena · ArXivMath · June 2026 | 85.4 | 91 · max |
| MathArena · ArXivMath · May 2026 | 87.5 | 87.5 · max |
| MathArena · BrokenArXiv · June 2026 | 47.8 | 86.6 · low |
| MathArena · BrokenArXiv · May 2026 | 44.5 | 78 · low |
| Vals LegalBench | 88.6 | 88.5 |
| Vals MedCode | 56.1 | 53.5 |
| Vals MedScribe | 88.5 | 91.3 |
| Vals Legal Research Bench | 49.5 | 55.3 |
| IFBench (instructions) | 63.5 | — |
| EQ-Bench 3 | 2,049.7 | — |
| MathArena · ArXivMath · August 2026 | — | 87.7 · high |
| MathArena · BrokenArXiv · April 2026 | 54.1 | — |
| MathArena · BrokenArXiv · August 2026 | — | 79.8 · high |
| MathArena · ArXivLean · June 2026 | 15.2 | — |
Specs
| Spec | Fable 5 | Fable 5.1 |
|---|---|---|
| Input, per 1M | $10 | $10 |
| Output, per 1M | $50 | $50 |
| Cost per task? | $8.75 | $2.98 |
| Speed, tok/s | 67 | 56 |
| Context | 1M | 1M |
| Weights | Closed | Closed |
| Released | Jun 2026 | Sep 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | Partial |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | Partial |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ? |
| Run code | ? | ? |
| Use a computer | ? | ? |
What breaks when you switch
Fable 5 → Fable 5.1
Replace forced tool_choice›
Forced tool_choice returns HTTP 400; use auto with an explicit instruction and strict tools, or JSON outputs.
Adaptive thinking is always on›
Disabled thinking or manual budget_tokens returns HTTP 400; omit thinking or use adaptive thinking.
No non-default sampling values›
Non-default temperature, top_p or top_k returns HTTP 400; omit non-default sampling values.
Requires 30-day data retention · Claude API›
A Claude API organization or workspace without 30-day retention receives HTTP 400; configure workspace retention or contact the Anthropic account team.
Thinking blocks bind to conversations›
Where conversation binding is enforced, replaying an invalidated block returns HTTP 400; use the documented drop_block control to continue without it.
No assistant prefill›
Assistant prefill returns HTTP 400; use system prompt instructions instead.
Fable 5.1 → Fable 5
Nothing known.