Claude Fable 5.1
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Math96.8
- 2Knowledge#295.4
- 3Coding#391.2
- 4Reasoning#585.2
- 5Agentic & tool use#259.8
Benchmark results (31)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Agentic & tool use
| MazeBench | 10 | #8 of 42 |
| AutomationBench | 31.4 | #19 of 25 |
| OSWorld 2.0 | 41.7 | #13 of 24 |
| Toolathlon-Verified | 77.8 | #3 of 22 |
| Terminal-Bench 4.0 | 55.80 | #6 of 18 |
| ApprenticeBench | 72 | #1 of 18 |
| CWE-bench v1 | 58.0 | #5 of 11 |
| Terminal-Bench-Science 0.1 | 52.6 | #5 of 5 |
| Toolathlon Verified Pass@3 | 81.5 | #4 of 4 |
| Toolathlon Verified Pass³ | 73.1 | #1 of 4 |
| Toolathlon Verified avg. turns(lower is better) | 23.7 | #2 of 4 |
Coding
| SWE-bench Pro | 81.2 | #3 of 69 |
| DeepSWE | 67.4 | #25 of 50 |
| Bug Hunt Bench | 43.0 | #4 of 23 |
| FrontierSWE v2 | 56.3 | #4 of 20 |
| ProgramBench | 87.6 | #3 of 15 |
| PostTrainBench v1.1 | 40.2 | #4 of 15 |
Multimodal
| Design Arena Website | 1314 | #9 of 91 |
| ParseBench | 78.92 | #2 of 37 |
| SWE-bench Multimodal | 54.7 | #4 of 7 |
Reasoning
| ARC-AGI-2 | 90.0 | #7 of 41 |
| ARC-AGI-1 | 97.50 | #4 of 36 |
| GraphWalks BFS 256K–1M | 65.0 | #4 of 4 |
Knowledge
| HLE | 65.0 | #1 of 69 |
| HLE w/o tools | 60.9 | #2 of 43 |
Math
| FrontierMath v2 (Tiers 1-3) | 90.175 | #4 of 64 |
| FrontierMath v2 (Tier 4) | 87.805 | #6 of 51 |
Multilingual
| SWE Multilingual | 89.1 | #5 of 46 |
Instruction following
| Gray Swan IPI (15 attempts)(lower is better) | 1.0 | #2 of 12 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| ✓ Official (Anthropic) checked Oct 17, 2026 · Claude Fable 5.1 | $10.00 | $50.00 | $0.25 |
| BenchLM list price | $10.00 | $50.00 | $0.25 |
| OpenRouter anthropic/claude-fable-5.1 | $10.00 | $50.00 | $0.25 |
Anthropic's September 2026 launch page lists `claude-fable-5-1` at $10 input / $50 output per million tokens and cuts cache reads to $0.25 per million tokens. Anthropic estimates that lower cache-read rate reduces typical usage-based Fable workloads by about 25% and highly agentic workloads by as much as 45%; cache-write and Batch rates are not inferred from those estimates.
Arena ratings
Human-preference ratings from LMArena. Best-rated variant shown.
| Text(Max) | 1512 | #3 · 14.6K votes |
| WebDev (coding)(Max) | 1743 | #5 · 7.7K votes |
| Vision(Max) | 1320 | #4 · 5K votes |
| Search(high) | 1246 | #5 · 12.8K votes |
| Documents(max) | 1513 | #2 · 1.4K votes |
Speed
Measured by us through OpenRouter (median of trials).
Not measured yet.
Related models
Same family
Similar score
- Gemini 4 Argon82.9 · $4.00
- GPT-6 Astra84.0 · $20.00
- GPT-6.1 Sol84.2 · $4.00
- Claude Sonnet 5.582.2 · $4.00
- Claude Fable 581.3 · $20.00