Mistral AIAuto-detectedMistral Medium (latest)
Best configuration ranks #382 of 610 on the BenchLeader Index at 45.8 ±4.4 (medium reasoning effort). Last measured 2 Sept 2026. Released 28 Apr 2026.
- Blended price
- $3.00/M
- $1.50 in · $7.50 out
- Output speed
- 133 tok/s
- First answer
- 2.30 s
- first token 2.30 s
- Context
- 262k
- Overall index46
- Reasoning44
- Coding38
- Agents & tools50
- Maths46
- Knowledge40
- Human preference59
- Multimodal50
- Composite36
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 |
|---|---|---|---|---|---|---|
| mediumbest | 45.8 | #382 | 133 tok/s | 2.30 s | $0.0029 | Agents & tools 50 · Coding 38 · Composite 36 · Human preference 59 · Knowledge 40 · Maths 46 · Multimodal 50 · Reasoning 44 |
| default | – | – | – | – | $0.0029 | Knowledge 41 |
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 | medium | default | Source |
|---|---|---|---|
| GPQA Diamond | 59.5%#179 | – | Epoch AI Benchmarking Hub |
| Humanity's Last Exam | 4.5%#40 | – | Scale AI / CAIS |
| LMArena Hard Prompts | 1427#121 | – | LMArena |
| GPQA Diamond (AA)not in index | 35.0%#466 | – | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 3.5%#505 | – | Artificial Analysis |
| Kagi LLM Benchmark | 50.0%#84 | – | Kagi LLM Benchmark |
Coding
| Benchmark | medium | default | Source |
|---|---|---|---|
| SciCode | 39.6%#113 | – | SciCode |
| WeirdML | 43.7%#85 | – | WeirdML |
| FrontierCode | 8.0%#29 | – | Cognition |
| LMArena Coding | 1454#123 | – | LMArena |
| LiveCodeBench | 44.9%#116 | – | Vals AI |
Agents & tools
| Benchmark | medium | default | Source |
|---|---|---|---|
| BFCL Overall | 37.7%#37 | – | Berkeley Function Calling Leaderboard |
Maths
| Benchmark | medium | default | Source |
|---|---|---|---|
| OTIS Mock AIME | 32.2%#186 | – | Epoch AI Benchmarking Hub |
| MATH Level 5 | 81.6%#33 | – | Epoch AI Benchmarking Hub |
| ProofBench | 9.0%#50 | – | Vals AI |
| AIME (Vals) | 42.3%#65 | – | Vals AI |
| MGSM | 90.9%#32 | – | Vals AI |
Knowledge
| Benchmark | medium | default | Source |
|---|---|---|---|
| MMLU-Pro | 75.3%#106 | – | Vals AI |
| LegalBench | 62.0%#127 | – | Vals AI |
| CorpFin | 50.7%#102 | – | Vals AI |
| TaxEval | 70.3%#89 | – | Vals AI |
| MedQA | 78.2%#72 | – | Vals AI |
| PRBench Finance | – | 39.4%#22 | Scale AI SEAL |
| PRBench Legal | – | 39.5%#25 | Scale AI SEAL |
Human preference
| Benchmark | medium | default | Source |
|---|---|---|---|
| LMArena Text | 1409#123 | – | LMArena |
Multimodal
| Benchmark | medium | default | Source |
|---|---|---|---|
| LMArena Vision | 1172#88 | – | LMArena |
Composite
| Benchmark | medium | default | Source |
|---|---|---|---|
| AA Intelligence Index | 5.5#488 | – | Artificial Analysis |
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 | $0.0029 | – | 4.6 s |
| Summarise a 30-page report | 12,000 / 600 | $0.022 | – | 6.8 s |
| Code edit | 6,000 / 1,500 | $0.020 | – | 13.6 s |
| Agentic coding session | 60,000 / 4,000 | $0.120 | – | 32.3 s |
| Structured extraction | 2,000 / 200 | $0.0045 | – | 3.8 s |
See also
Data as of 9 Sept 2026. Compare these configurations.