Mistral AIAuto-detectedMistral Small (latest)
Best configuration ranks #492 of 610 on the BenchLeader Index at 41.1 ±4.9. Last measured 2 Sept 2026. Released 20 Jun 2025.
- Blended price
- $0.262/M
- $0.150 in · $0.600 out
- Output speed
- 155 tok/s
- First answer
- 0.79 s
- first token 0.43 s
- Context
- 256k
- Overall index41
- Reasoning41
- Coding29
- Agents & tools49
- Maths34
- Knowledge30
- Human preference53
- Multimodal46
- Composite36
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 | default | Source |
|---|---|---|
| GPQA Diamond | 47.5%#208 | Epoch AI Benchmarking Hub |
| LMArena Hard Prompts | 1374#171 | LMArena |
| GPQA Diamond (AA)not in index | 38.1%#459 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 4.3%#440 | Artificial Analysis |
| GPQA Diamond (Vals)not in index | 50.8%#114 | Vals AI |
| Kagi LLM Benchmark | 37.8%#109 | Kagi LLM Benchmark |
Coding
| Benchmark | default | Source |
|---|---|---|
| SciCode | 26.5%#146 | SciCode |
| LMArena Coding | 1412#163 | LMArena |
| LiveCodeBench | 31.8%#131 | Vals AI |
Agents & tools
| Benchmark | default | Source |
|---|---|---|
| BFCL Overall | 37.1%#39 | Berkeley Function Calling Leaderboard |
Maths
| Benchmark | default | Source |
|---|---|---|
| OTIS Mock AIME | 5.8%#224 | Epoch AI Benchmarking Hub |
| MATH Level 5 | 46.8%#54 | Epoch AI Benchmarking Hub |
| AIME (Vals) | 5.6%#88 | Vals AI |
| MGSM | 84.2%#66 | Vals AI |
Knowledge
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1357#172 | LMArena |
Multimodal
| Benchmark | default | Source |
|---|---|---|
| LMArena Vision | 1141#100 | LMArena |
Composite
| Benchmark | default | Source |
|---|---|---|
| AA Intelligence Index | 5.8#468 | Artificial Analysis |
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 |
|---|---|---|---|---|---|---|
| Mistral | 108 tok/s | 0.45 s | $0.150 | $0.600 | 262k | – |
| Mistral (ZDR) | 99 tok/s | 0.42 s | $0.150 | $0.600 | 262k | – |
| Venice | 99 tok/s | 0.57 s | $0.188 | $0.750 | 256k | fp8 |
| Mistral (EU) | 25 tok/s | 0.39 s | $0.165 | $0.660 | 262k | – |
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. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0002 | – | 2.7 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0022 | – | 4.7 s |
| Code edit | 6,000 / 1,500 | $0.0018 | – | 10.5 s |
| Agentic coding session | 60,000 / 4,000 | $0.011 | – | 26.6 s |
| Structured extraction | 2,000 / 200 | $0.0004 | – | 2.1 s |
See also
Data as of 9 Sept 2026. Compare with another model.