AnthropicClaude Sonnet 4.6
Best configuration ranks #89 of 610 on the BenchLeader Index at 60.3 ±3.2. Last measured 8 Sept 2026. Released 17 Feb 2026.
- Blended price
- $6.00/M
- $3.00 in · $15.00 out
- Output speed
- 43 tok/s
- First answer
- 2.12 s
- first token 1.47 s
- Context
- 1M
- Overall index60
- Reasoning65
- Coding57
- Agents & tools55
- Maths63
- Knowledge62
- Instruction following57
- Human preference62
- Multimodal61
- Long context67
- Composite67
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 | 55.6 | #174 | 43 tok/s | 1.77 s | $0.0057 | Agents & tools 71 · Composite 58 · Instruction following 44 · Knowledge 63 · Long context 61 · Multimodal 54 |
| medium | 51.8 | #266 | 43 tok/s | 2.12 s | $0.0057 | Agents & tools 38 · Coding 61 · Composite 45 · Maths 60 · Reasoning 61 |
| high | 52.3 | #252 | 43 tok/s | 2.12 s | $0.0057 | Agents & tools 49 · Knowledge 46 · Maths 57 · Reasoning 61 |
| max | 49.7 | #302 | 43 tok/s | 2.12 s | $0.0057 | Agents & tools 36 · Coding 53 · Knowledge 44 · Maths 55 · Reasoning 60 |
| defaultbest | 60.3 | #89 | 43 tok/s | 2.12 s | $0.0057 | Agents & tools 55 · Coding 57 · Composite 67 · Human preference 62 · Instruction following 57 · Knowledge 62 · Long context 67 · Maths 63 · Multimodal 61 · Reasoning 65 |
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 | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| GPQA Diamond | – | 83.3%#91 | 83.3%#91 | 78.8%#120 | 87.4%#61 | Epoch AI Benchmarking Hub | |
| LMArena Hard Prompts | – | – | – | – | 1504#19 | LMArena | |
| LiveBench Reasoningnot in index | – | 84.8%#30 | – | – | – | LiveBench | |
| GPQA Diamond (AA)not in index | 79.7%#190 | – | – | – | 87.5%#96 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 11.2%#233 | – | – | – | 33.6%#89 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | – | – | – | 85.6%#44 | Vals AI | |
| ARC-AGI-1 | – | – | 86.5%#67 | 86.0%#70 | – | ARC Prize | |
| ARC-AGI-2 | – | – | 60.4%#56 | 58.3%#62 | – | ARC Prize |
Coding
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| SWE-bench Verified (Epoch) | – | – | – | – | 75.2%#14 | Epoch AI Benchmarking Hub | |
| SciCode | – | – | – | 46.8%#79 | – | SciCode | |
| WeirdML | – | 66.1%#37 | – | – | – | WeirdML | |
| FrontierCode | – | – | – | – | 24.3%#23 | Cognition | |
| LMArena Coding | – | – | – | – | 1528#15 | LMArena | |
| LMArena WebDev | – | – | – | – | 1521#32 | LMArena | |
| LiveBench Codingnot in index | – | 79.3%#16 | – | – | – | LiveBench | |
| SciCode (AA)not in index | – | – | – | – | 50.1%#68 | Artificial Analysis | |
| LiveCodeBench | – | – | – | – | 82.1%#55 | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | – | 77.4%#34 | Vals AI |
Agents & tools
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | 53.4%#22 | Terminal-Bench | |
| OSWorld-Verified 2.0 | – | 9.3%#10 | – | 8.3%#11 | – | OSWorld | |
| APEX-Agents | – | – | 23.7%#32 | – | – | Mercor | |
| LMArena Agent | – | – | – | – | -1#27 | LMArena | |
| LiveBench Agentic Codingnot in index | – | 42.6%#45 | – | – | – | LiveBench | |
| Terminal-Bench Hard | 42.4%#44 | – | – | – | 53.0%#17 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 79.0%#125 | – | – | – | 79.5%#122 | Artificial Analysis | |
| Terminal-Bench 2.1 (Vals) | – | – | – | – | 57.3%#34 | Vals AI | |
| MCP Atlas | – | – | – | – | 69.5%#20 | Scale AI SEAL |
Maths
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| OTIS Mock AIME | – | 82.2%#93 | 75.6%#111 | 71.1%#120 | 85.8%#80 | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | 45.0%#22 | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | 87.0%#35 | – | – | – | LiveBench | |
| AIME (Vals) | – | – | – | – | 92.3%#20 | Vals AI |
Knowledge
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | 35.5%#41 | 32.8%#51 | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | 78.0%#21 | – | – | – | LiveBench | |
| AA-Omniscience | -2.1#104 | – | – | – | 12.2#60 | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | 87.3%#29 | Vals AI | |
| LegalBench | – | – | – | – | 82.1%#67 | Vals AI | |
| CorpFin | – | – | – | – | 65.3%#29 | Vals AI | |
| TaxEval | – | – | – | – | 77.1%#4 | Vals AI | |
| MedQA | – | – | – | – | 92.1%#36 | Vals AI |
Instruction following
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | 76.1%#36 | – | – | – | LiveBench | |
| IFBench | 42.4%#233 | – | – | – | 56.6%#139 | Artificial Analysis |
Human preference
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | – | 1472#35 | LMArena | |
| EQ-Bench 4 | – | – | – | – | 1207#15 | EQ-Bench |
Multimodal
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | – | 1282#25 | LMArena | |
| MMMU-Pro | 69.2%#124 | – | – | – | 73.3%#96 | Artificial Analysis |
Long context
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| AA-LCR | 69.3%#170 | – | – | – | 80.0%#51 | Artificial Analysis |
Composite
| Benchmark | low | medium | high | max | default | Source | Trend |
|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | 152.3#34 | Epoch AI Benchmarking Hub | |
| LiveBench | – | 73.0%#38 | – | – | – | LiveBench | |
| AA Intelligence Index | 23.3#133 | – | – | – | 30.4#77 | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | – | 50.6#28 | 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 |
|---|---|---|---|---|---|---|
| Amazon Bedrock (EU) | 69 tok/s | 1.17 s | $3.30 | $16.50 | 1M | – |
| Claude Platform on AWS | 38 tok/s | 1.21 s | $3.00 | $15.00 | 1M | – |
| Google Vertex (Global) | 37 tok/s | 1.95 s | $3.00 | $15.00 | 1M | – |
| Azure | 36 tok/s | 1.97 s | $3.00 | $15.00 | 1M | – |
| Anthropic | 32 tok/s | 1.73 s | $3.00 | $15.00 | 1M | – |
| Amazon Bedrock (Global) | 29 tok/s | 1.04 s | $3.00 | $15.00 | 1M | – |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
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 $0.300 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0057 | $0.0049 | 9.0 s |
| Summarise a 30-page report | 12,000 / 600 | $0.045 | $0.021 | 15.9 s |
| Code edit | 6,000 / 1,500 | $0.041 | $0.028 | 36.6 s |
| Agentic coding session | 60,000 / 4,000 | $0.240 | $0.118 | 1.6 min |
| Structured extraction | 2,000 / 200 | $0.0090 | $0.0050 | 6.7 s |
See also
Data as of 9 Sept 2026. Compare these configurations.