AnthropicReasoning modelClaude Fable 5.1
Best configuration ranks #3 of 610 on the BenchLeader Index at 70.3 ±7.1 (xhigh reasoning effort). Last measured 1 Sept 2026. Released 1 Sept 2026.
- Blended price
- $20.00/M
- $10.00 in · $50.00 out
- Output speed
- 58 tok/s
- First answer
- 132 s
- first token 6.84 s
- Context
- 1M
- Overall index70
- Reasoning76
- Coding71
- Knowledge84
- Long context68
- Composite95
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| low | 65.9 | #22 | 49 tok/s | 6.78 s | $0.019 | Coding 65 · Composite 88 · Knowledge 80 · Long context 68 · Reasoning 68 |
| medium | 68.0 | #13 | 52 tok/s | 12 s | $0.019 | Coding 66 · Composite 91 · Knowledge 82 · Long context 69 · Reasoning 71 |
| high | 69.7 | #8 | 54 tok/s | 25 s | $0.019 | Agents & tools 72 · Coding 74 · Composite 94 · Knowledge 83 · Long context 69 · Reasoning 72 |
| xhighbest | 70.3 | #3 | 58 tok/s | 132 s | $0.019 | Coding 71 · Composite 95 · Knowledge 84 · Long context 68 · Reasoning 76 |
| max | 70.2 | #4 | 67 tok/s | 286 s | $0.019 | Agents & tools 67 · Coding 78 · Composite 77 · Human preference 71 · Knowledge 76 · Maths 77 · Reasoning 72 |
| default | 70.0 | #6 | 67 tok/s | 286 s | $0.019 | Agents & tools 72 · Coding 66 · Composite 95 · Knowledge 70 · Long context 69 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| Humanity's Last Exam | – | – | – | 46.5%#1 | – | – | Scale AI / CAIS | |
| LMArena Hard Prompts | – | – | – | – | 1522#5 | – | LMArena | |
| LiveBench Reasoningnot in index | – | – | – | – | 91.7%#2 | – | LiveBench | |
| GPQA Diamond (AA)not in index | 88.1%#91 | 88.6%#87 | 90.6%#58 | 93.4%#21 | – | 93.7%#11 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 48.9%#16 | 53.8%#9 | 55.9%#3 | 58.7%#2 | – | 59.1%#1 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | – | 93.4%#7 | Vals AI | |
| ARC-AGI-1 | 90.0%#53 | 94.5%#24 | 96.0%#18 | 96.5%#13 | 97.5%#6 | – | ARC Prize | |
| ARC-AGI-2 | 78.3%#27 | 86.3%#15 | 88.8%#11 | 90.0%#7 | 90.0%#7 | – | ARC Prize |
Coding
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| SciCode | 55.7%#21 | 55.3%#26 | 57.6%#9 | 60.1%#3 | 62.0%#1 | – | SciCode | |
| WeirdML | – | – | 92.3%#3 | – | 92.9%#1 | – | WeirdML | |
| FrontierCode | – | 50.9%#4 | – | – | – | – | Cognition | |
| LMArena Coding | – | – | – | – | 1517#34 | – | LMArena | |
| LMArena WebDev | – | – | – | – | 1764#2 | – | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | 86.4%#1 | – | LiveBench | |
| SciCode (AA)not in index | 56.7%#18 | 56.4%#22 | 58.7%#10 | 60.9%#3 | – | 63.1%#1 | Artificial Analysis | |
| LiveCodeBench | – | – | – | – | – | 90.5%#1 | Vals AI |
Agents & tools
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | 57.9%#16 | – | Terminal-Bench | |
| APEX-Agents | – | – | 44.4%#4 | – | – | 47.4%#1 | Mercor | |
| LMArena Agent | – | – | – | – | 14.5#1 | – | LMArena | |
| LiveBench Agentic Codingnot in index | – | – | – | – | 66.1%#1 | – | LiveBench | |
| Terminal-Bench 2.1 (Vals) | – | – | – | – | – | 85.0%#3 | Vals AI | |
| MCP Atlas | – | – | – | – | – | 87.2%#2 | Scale AI SEAL | |
| HiL-Bench | – | – | – | – | – | 61.5%#1 | Scale AI SEAL |
Maths
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | – | 90.2%#2 | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | – | 87.8%#6 | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | – | – | – | – | 100.0%#1 | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | 100.0%#1 | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | 97.0%#1 | – | LiveBench |
Knowledge
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | 70.8%#3 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | 80.3%#4 | – | LiveBench | |
| AA-Omniscience | 34.1#13 | 37.6#10 | 40.8#8 | 42.4#6 | – | 43.5#2 | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | – | 92.4%#1 | Vals AI | |
| LegalBench | – | – | – | – | – | 88.5%#2 | Vals AI | |
| TaxEval | – | – | – | – | – | 76.0%#8 | Vals AI | |
| PRBench Finance | – | – | – | – | – | 50.8%#7 | Scale AI SEAL | |
| PRBench Legal | – | – | – | – | – | 51.6%#5 | Scale AI SEAL |
Instruction following
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | 89.5%#2 | – | LiveBench |
Human preference
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | – | 1504#3 | – | LMArena |
Long context
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| AA-LCR | 82.3%#21 | 84.7%#3 | 83.7%#8 | 83.0%#12 | – | 85.3%#2 | Artificial Analysis |
Composite
| Benchmark | low | medium | high | xhigh | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | 164.2#2 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | 83.4%#1 | – | LiveBench | |
| AA Intelligence Index | 47.0#15 | 49.1#11 | 51.2#5 | 53.2#2 | – | 53.4#1 | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | – | – | 68.8#1 | Vals AI |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Google Vertex | 49 tok/s | 4.60 s | $10.00 | $50.00 | 1M | – |
| Anthropic | 41 tok/s | 6.84 s | $10.00 | $50.00 | 1M | – |
| Azure | 21 tok/s | 8.05 s | $10.00 | $50.00 | 1M | – |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.250 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.019 | $0.016 | 2.3 min |
| Summarise a 30-page report | 12,000 / 600 | $0.150 | $0.062 | 2.4 min |
| Code edit | 6,000 / 1,500 | $0.135 | $0.091 | 2.6 min |
| Agentic coding session | 60,000 / 4,000 | $0.800 | $0.361 | 3.3 min |
| Structured extraction | 2,000 / 200 | $0.030 | $0.015 | 2.3 min |
See also
Data as of 9 Sept 2026. Compare these configurations.