Mistral Small 4.
Mistral AI Mar 2026 Open weights Small
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AA-LCR (Artificial Analysis Long Context Reasoning) | 71.2% | #6 of 38 | Lab model card | |
| MMLU-Pro | 78% | #93 of 167 | Lab model card | |
| AraGen | 33.9% | #28 of 42 | Official leaderboard | |
| COLLIE | 62.9% | #8 of 12 | Lab model card | |
| Arena-Hard-Auto | 58.3% | #30 of 39 | Lab model card | |
| Longform Creative Writing (EQ-Bench) | 43.2 | #94 of 119 | Official leaderboard | |
| Last Translation Benchmark | 1.32 | #29 of 36 | Official leaderboard | |
| IFBench | 48% | #65 of 80 | Lab model card | |
| SEA-HELM | 52.1 | #59 of 71 | Official leaderboard | |
| Judgemark v2.1 | 51.6 | #38 of 42 | Official leaderboard | |
| BullshitBench V2 | 6.00% | #123 of 129 | Official leaderboard | |
| Judgemark v4 | 16.5 | #39 of 40 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AIME 2025 | 83.8% | #77 of 143 | Lab model card | |
| GPQA Diamond | 71.2% | #177 of 302 | Lab model card | |
| DTBench | 70.9% | #112 of 180 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LiveCodeBench | 63.6% | #21 of 54 | Lab model card | |
| StructEval | 72.5% | #17 of 28 | Official leaderboard | |
| FDE-Gym | -0.32 | #11 of 12 | Official leaderboard | |
| ALE-Bench | 862 | #117 of 126 | Official leaderboard | |
| Vibe Code Bench | 0% | #119 of 120 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| PinchBench | 71.3% | #20 of 48 | Official leaderboard | |
| PinchBench 2.0 | 64.6% | #39 of 57 | Official leaderboard | |
| GAIA | 17.6% | #56 of 64 | Official leaderboard | |
| Pencil Puzzle Bench | 0% | #48 of 53 | Official leaderboard | |
| Clocktower Radio | 1,242 | #22 of 23 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| CorpBench Work | 74.8 | #13 of 17 | Official leaderboard | |
| Vals Index v1 | 29.1% | #33 of 34 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MMMU-Pro | 60% | #67 of 90 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Human Pathogen Capabilities Test v2 (HPCT v2) | 27.5% | #67 of 71 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Conceptual Reasoning Index | 30.5 | #115 of 151 | Official leaderboard | |
| LMCA | 20.6% | #112 of 147 | Official leaderboard | |
| ACCoRD | 39.0 | #134 of 152 | Official leaderboard | |
| Political Science LLM Benchmark | 62.9% | #39 of 42 | Official leaderboard | |
| SAGE v1.1 (Vals) | 59.2% | #29 of 29 | Official leaderboard |
The Self-Reported Index fits one score to the 9 capability results Mistral AI reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 130.6 to 138.5. How we compute it.
The benchmarks the lab chose to headline, with the scores it gave.