Mistral AIMistral Medium 3
Mistral Medium 3 is a Mistral AI proprietary model, released 7 May 2025. Its best configuration ranks #546 of 785 on the BenchLeader Index at 44.1 ±3.4, in the lower half. The ± is the point: 284 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in human preference (56) and lowest in coding (39). At $0.800 per million tokens blended it is mid-priced. Output speed of 47 tokens per second puts it in the slowest quarter, with a first answer in 2.4 s. Last measured 22 Oct 2026.
- Blended price
- $0.800/M
- $0.400 in · $2.00 out
- Output speed
- 47 tok/s
- measured by Artificial Analysis
- First answer
- 2.41 s
- first token 0.39 s
- Context
- 131k
- Full answer
- 13 s
- median, reasoning included
- Cost per run
- –
- Released
- 7 May 2025
- Overall index44
- Reasoning42
- Coding39
- Agents & tools45
- Maths50
- Knowledge42
- Instruction following42
- Human preference56
- Multimodal40
- Long context40
- Composite40
Versions
Mistral AI has shipped 4 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| Mistral Medium 3.5 | 28 Apr 2026 | 49.4 | #397 |
| Mistral Medium 3.1 | 12 Aug 2025 | 46.1 | #494 |
| Mistral Medium 3this page | 7 May 2025 | 44.1 | #546 |
| Mistral Medium 2312 | 11 Dec 2023 | – | – |
How its index has moved
9 Sept 2026 to 21 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. 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 | not stated | Source | Trend |
|---|---|---|---|
| GPQA Diamond | 59.5%#193 | Epoch AI Benchmarking Hub | |
| Humanity's Last Exam | 4.5%#44 | Scale AI / CAIS | |
| Arena Hard Prompts | 1401#173 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 57.8%#389 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 4.0%#547 | Artificial Analysis | |
| CritPt | 0.0%#315 | Artificial Analysis | |
| DTBenchnot in index | 62.3%#176 | Epoch AI Benchmarking Hub |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Coding | 1432#170 | Arena (LMArena) | |
| LiveCodeBench | 44.9%#116 | Vals AI |
Agents & tools
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Terminal-Bench Hard | 3.8%#298 | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 24.3%#302 | Artificial Analysis | |
| BFCL Overall | 37.7%#37 | Berkeley Function Calling Leaderboard |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| OTIS Mock AIME | 32.2%#204 | Epoch AI Benchmarking Hub | |
| MATH Level 5 | 81.6%#34 | Epoch AI Benchmarking Hub | |
| AIME (Vals) | 42.3%#64 | Vals AI | |
| MGSM | 90.9%#31 | Vals AI | |
| MATH 500not in index | 87.0%#32 | Vals AI | |
| Arena Maths | 1348#200 | Arena (LMArena) |
Knowledge
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-Omniscience | -31.4#286 | Artificial Analysis | |
| MMLU-Pro | 75.3%#105 | Vals AI | |
| LegalBench | 62.0%#133 | Vals AI | |
| CorpFin | 50.7%#101 | Vals AI | |
| TaxEval | 70.3%#88 | Vals AI | |
| MedQA | 78.2%#71 | Vals AI | |
| AA-Omniscience: accuracynot in index | 18.3%#339 | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 39.1%#165 | Artificial Analysis |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| IFBench | 39.3%#269 | Artificial Analysis | |
| Arena Instruction Followingnot in index | 1366#177 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1387#167 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1364#155 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1403#151 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1392#167 | Arena (LMArena) | |
| Lech Mazur Creative Writingnot in index | 77.3%#19 | Lech Mazur |
Multimodal
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Vision | 1157#114 | Arena (LMArena) | |
| MMMU-Pro | 53.0%#233 | Artificial Analysis | |
| VISTA | 34.6%#45 | Scale AI SEAL | |
| MMMU-Pro (Vals)not in index | 63.0%#79 | Vals AI | |
| MortgageTaxnot in index | 36.5%#88 | Vals AI |
Long context
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| AA-LCR | 31.3%#391 | Artificial Analysis |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 134.0#136 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 9.1#411 | Artificial Analysis |
Safety & honesty
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| MASKnot in index | 42.6%#62 | Scale AI SEAL |
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 | 106 tok/s | 0.55 s | $0.400 | $2.00 | 131k | – |
| Mistral | 86 tok/s | 0.30 s | $0.400 | $2.00 | 131k | – |
Price history
The price above, per 1M tokens, every day it changed.
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 | $0.0008 | – | 8.8 s |
| Summarize a 30-page report | 12,000 / 600 | $0.0060 | – | 15.3 s |
| Code edit | 6,000 / 1,500 | $0.0054 | – | 34.6 s |
| Agentic coding session | 60,000 / 4,000 | $0.032 | – | 1.5 min |
| Structured extraction | 2,000 / 200 | $0.0012 | – | 6.7 s |
More from Mistral AI
See also
Questions people ask
- What are Mistral Medium 3's benchmark scores?
- As of 25 Oct 2026, Mistral Medium 3 scores 44.1 ±3.4 on the BenchLeader Index, strongest in human preference (56) and weakest in coding (39). That rests on 26 independent benchmark results, and every score on this page links to its publisher.
- How does Mistral Medium 3 rank?
- Mistral Medium 3 is #546 of 785 ranked configurations on 25 Oct 2026. 284 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Mistral Medium 3 cost per 1M tokens?
- As of 25 Oct 2026, Mistral Medium 3 costs $0.400 per 1M input tokens and $2.00 per 1M output tokens, $0.800 blended (three input tokens per output token). The list price last changed on 9 Sept 2026.
- When was Mistral Medium 3 released?
- Mistral AI released Mistral Medium 3 on 7 May 2025.
Data as of 25 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Mistral Medium 3: benchmarks, pricing, speed and rank”, data as of 25 Oct 2026, https://www.benchleader.com/models/mistral-medium-3.