Claude Fable 5.1 vs Claude Opus 5.5
Humanity's Last Exam Humanity's Last Exam: Very hard expert questions across many fields, written to stump today’s models
Fable 5.153.8
Opus 5.554.7
Arena Text · overall Arena Text · overall: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,502
Opus 5.5 · high1,508
Arena Text · Chinese Arena Text · chinese: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,585.6
Opus 5.5 · high1,583.6
Arena Text · English Arena Text · english: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,504.9
Opus 5.5 · high1,516.9
Arena Text · French Arena Text · french: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,527.3
Opus 5.5 · high1,505.3
Arena Text · instruction following Arena Text · instruction following: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,517.2
Opus 5.5 · high1,535.8
Arena Text · non-English Arena Text · non english: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,507.5
Opus 5.5 · high1,506.9
Arena Text · Russian Arena Text · russian: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,518.6
Opus 5.5 · high1,519.1
Arena Text · Spanish Arena Text · spanish: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,520.5
Opus 5.5 · high1,519
AA Omniscience Index AA Omniscience Index: Scores factual knowledge while penalizing incorrect answers to encourage admitting uncertainty
Fable 5.137.6
Opus 5.540.3
SimpleQA Verified v1 SimpleQA Verified v1: Accuracy on short factual questions with verified answers, answered without web access
Fable 5.1 · max70.8
Opus 5.5 · max72.2
CritPt (physics) CritPt (physics): Research-level physics problems that need many careful reasoning steps
Fable 5.129.1
Opus 5.527.7
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 5.148.9
Opus 5.551.2
AA Omniscience · Accuracy AA Omniscience · Accuracy: Measures correct answers on the AA Omniscience factual knowledge test
Fable 5.163.1
Opus 5.564.6
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 5.169.1
Opus 5.568.4
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 5.184.7
Opus 5.584.3
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 5.126.8
Opus 5.525.6
MLCR-AA MLCR-AA: Models synthesize long medical records; the headline credits concise answers judged both accurate and complete
Fable 5.1 · max71.1
Opus 5.5 · max66.7
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 5.196.6
Opus 5.598.3
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 5.194.5
Opus 5.597.5
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 5.187.5
Opus 5.593.5
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 5.186.3
Opus 5.587.5
Chess Puzzles Chess Puzzles: Models choose the best next move from a novel chess position
Fable 5.1 · max48
Opus 5.5 · max67
FrontierMath Erdos FrontierMath Erdos: Models solve expert-written advanced mathematics problems on the named tier and edition
Fable 5.1 · max0
Opus 5.5 · max2.9
FrontierMath Tier 4 (v2) FrontierMath Tier 4 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Fable 5.1 · max87.8
Opus 5.5 · max95
FrontierMath Tiers 1–3 (v2) FrontierMath Tiers 1–3 (v2): Models solve expert-written advanced mathematics problems on the named tier and edition
Fable 5.1 · max90.2
Opus 5.5 · max91.2
Mystery Game Puzzles Mystery Game Puzzles: Models choose the best move in puzzle positions from an undisclosed game
Fable 5.1 · max58
Opus 5.5 · max71
MathArena · ArXivMath · August 2026 MathArena · arxiv · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · high87.7
Opus 5.5 · high86
MathArena · ArXivMath · June 2026 MathArena · arxiv · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · max91
Opus 5.5 · high83.3
MathArena · ArXivMath · May 2026 MathArena · arxiv · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · max87.5
Opus 5.5 · high81.7
MathArena · BrokenArXiv · August 2026 MathArena · arxiv_false · 08/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · high79.8
Opus 5.5 · high77.4
MathArena · BrokenArXiv · June 2026 MathArena · arxiv_false · 06/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · low86.6
Opus 5.5 · high100
MathArena · BrokenArXiv · May 2026 MathArena · arxiv_false · 05/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · low78
Opus 5.5 · high95
Vals MedCode Vals MedCode: Medical coding accuracy on patient encounters
Fable 5.153.5
Opus 5.549.8
Vals MedScribe Vals MedScribe: Accuracy of clinical notes produced from consultation transcripts
Fable 5.191.3
Opus 5.591.4
Vals Legal Research Bench Vals Legal Research Bench: Grounded legal research; a task passes only if all criteria pass
Fable 5.155.3
Opus 5.550.5
Arena Text · German Arena Text · german: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,501.7
Opus 5.5—
Arena Text · Japanese Arena Text · japanese: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,560.1
Opus 5.5—
Arena Text · Korean Arena Text · korean: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,520.8
Opus 5.5—
Arena Text · Polish Arena Text · polish: Human pairwise preferences in the stated text vote pool
Fable 5.1 · max1,524.3
Opus 5.5—
MathArena · ArXivMath · April 2026 MathArena · arxiv · 04/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1 · max77.5
Opus 5.5—
MathArena · ArXivMath · September 2026 MathArena · arxiv · 09/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1—
Opus 5.5 · high82.1
MathArena · BrokenArXiv · September 2026 MathArena · arxiv_false · 09/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1—
Opus 5.5 · high76.7
MathArena · ArXivLean · September 2026 MathArena · arxivlean · 09/2026: Answer accuracy on the named mathematics competition under the recorded effort
Fable 5.1—
Opus 5.5 · high47.2
Vals LegalBench Vals LegalBench: Legal reasoning and legal-language understanding on LegalBench
Fable 5.188.5
Opus 5.5—
Vals Science Index Vals Science Index: Scientific problem solving across the index’s component evaluations
Fable 5.1—
Opus 5.558.7
| Benchmark | Fable 5.1 | Opus 5.5 |
|---|---|---|
| Humanity's Last Exam | 53.8 | 54.7 |
| Arena Text · overall | 1,502 · max | 1,508 · high |
| Arena Text · Chinese | 1,585.6 · max | 1,583.6 · high |
| Arena Text · English | 1,504.9 · max | 1,516.9 · high |
| Arena Text · French | 1,527.3 · max | 1,505.3 · high |
| Arena Text · instruction following | 1,517.2 · max | 1,535.8 · high |
| Arena Text · non-English | 1,507.5 · max | 1,506.9 · high |
| Arena Text · Russian | 1,518.6 · max | 1,519.1 · high |
| Arena Text · Spanish | 1,520.5 · max | 1,519 · high |
| AA Omniscience Index | 37.6 | 40.3 |
| SimpleQA Verified v1 | 70.8 · max | 72.2 · max |
| CritPt (physics) | 29.1 | 27.7 |
| AA Intelligence Index 4.3.2 | 48.9 | 51.2 |
| AA Omniscience · Accuracy | 63.1 | 64.6 |
| AA Omniscience · Hallucination Rate | 69.1 | 68.4 |
| AA-LCR (long context) 1.1 | 84.7 | 84.3 |
| GDP.pdf · AA all-pass | 26.8 | 25.6 |
| MLCR-AA | 71.1 · max | 66.7 · max |
| ARC-AGI-1 · public evaluation | 96.6 | 98.3 |
| ARC-AGI-1 · semi-private | 94.5 | 97.5 |
| ARC-AGI-2 · public evaluation | 87.5 | 93.5 |
| ARC-AGI-2 · semi-private | 86.3 | 87.5 |
| Chess Puzzles | 48 · max | 67 · max |
| FrontierMath Erdos | 0 · max | 2.9 · max |
| FrontierMath Tier 4 (v2) | 87.8 · max | 95 · max |
| FrontierMath Tiers 1–3 (v2) | 90.2 · max | 91.2 · max |
| Mystery Game Puzzles | 58 · max | 71 · max |
| MathArena · ArXivMath · August 2026 | 87.7 · high | 86 · high |
| MathArena · ArXivMath · June 2026 | 91 · max | 83.3 · high |
| MathArena · ArXivMath · May 2026 | 87.5 · max | 81.7 · high |
| MathArena · BrokenArXiv · August 2026 | 79.8 · high | 77.4 · high |
| MathArena · BrokenArXiv · June 2026 | 86.6 · low | 100 · high |
| MathArena · BrokenArXiv · May 2026 | 78 · low | 95 · high |
| Vals MedCode | 53.5 | 49.8 |
| Vals MedScribe | 91.3 | 91.4 |
| Vals Legal Research Bench | 55.3 | 50.5 |
| Arena Text · German | 1,501.7 · max | — |
| Arena Text · Japanese | 1,560.1 · max | — |
| Arena Text · Korean | 1,520.8 · max | — |
| Arena Text · Polish | 1,524.3 · max | — |
| MathArena · ArXivMath · April 2026 | 77.5 · max | — |
| MathArena · ArXivMath · September 2026 | — | 82.1 · high |
| MathArena · BrokenArXiv · September 2026 | — | 76.7 · high |
| MathArena · ArXivLean · September 2026 | — | 47.2 · high |
| Vals LegalBench | 88.5 | — |
| Vals Science Index | — | 58.7 |
Specs
| Spec | Fable 5.1 | Opus 5.5 |
|---|---|---|
| Input, per 1M | $10 | $4 |
| Output, per 1M | $50 | $20 |
| Cost per task? | $2.98 | $1.34 |
| Speed, tok/s | 56 | 81 |
| Context | 1M | 1M |
| Weights | Closed | Closed |
| Released | Sep 2026 | Sep 2026 |
| See images | ✓ | ✓ |
| Read PDFs | Partial | ✓ |
| Hear audio | — | — |
| Watch video | — | — |
| Call tools | ✓ | ✓ |
| Force a tool call | Partial | — |
| Return strict JSON | ✓ | ✓ |
| Search the web | ? | ✓ |
| Run code | ? | ✓ |
| Use a computer | ? | ✓ |
What breaks when you switch
Fable 5.1 → Opus 5.5
No forced tool use›
Forced tool_choice any or tool returns HTTP 400; remove the forced choice.
Thinking is always on›
Disabled thinking or a manual budget returns HTTP 400; omit thinking or use adaptive thinking.
Remove sampling parameters›
Non-default temperature, top_p or top_k returns HTTP 400; omit these parameters and use prompting.
No assistant prefill›
Assistant prefill returns HTTP 400; use structured outputs or system instructions, with system instructions on Bedrock.
Old computer tool rejected · Claude API / Google Cloud›
The computer_20251124 tool returns HTTP 400 on the Claude API and Google Cloud; use computer_toolset_20260801 there.
Return thinking blocks unmodified›
Edited, reordered or partially dropped thinking blocks return HTTP 400; echo assistant messages complete and unmodified with tool results.
Opus 5.5 → 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.