Ministral-3-3B-Reasoning-2512: benchmark results
Mistral Ministral 3 3B Reasoning 2512 snapshot. Provider: Mistral. Released 2025-12-02. Access: Open.
Aggregate Rating 1460 ± 17, rank #932 of 1656 rated models, from 307 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Latvian NLU - MultiWikiQA LV | 68.05 | Character F1 (%) | 92.2 |
| EuroEval Spanish NLU - ScaLA ES | 40.83 | Matthews correlation coefficient (%) | 91.5 |
| EuroEval Italian NLU - ScaLA IT | 45.9 | Matthews correlation coefficient (%) | 90.8 |
| EuroEval Portuguese NLU - ScaLA PT | 42.57 | Matthews correlation coefficient (%) | 90.5 |
| EuroEval Catalan NLU - MultiWikiQA CA | 74.41 | Character F1 (%) | 90 |
| EuroEval Portuguese NLU - MultiWikiQA PT | 76.14 | Character F1 (%) | 89.4 |
| EuroEval Croatian NLU - Multi Wiki QA HR | 66.43 | Character F1 (%) | 89 |
| EuroEval Portuguese Summarization - Publico | 33.35 | chrF3++ (%) | 87.9 |
| EuroEval French NLU - ScaLA FR | 50.46 | Matthews correlation coefficient (%) | 87.2 |
| EuroEval Lithuanian NLU - MultiWikiQA LT | 68.2 | Character F1 (%) | 87.1 |
| EuroEval Romanian NLU - MultiWikiQA RO | 71.92 | Character F1 (%) | 86.7 |
| EuroEval Danish NLU - Multi Wiki QA DA | 78.06 | Character F1 (%) | 86.2 |
Interactive version: theaggregate.ai/model?slug=ministral-3-3b-reasoning-2512 · How It Works · Data refreshed daily, snapshot 2026-10-20.