Mistral Small 3.1: benchmark results
Mistral AI's Small 3.1, a 24B multimodal update to Small 3 (March 2025). Provider: Mistral. Released 2025-03-17. Access: Open.
Aggregate Rating 1551 ± 7, rank #610 of 1656 rated models, from 747 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EuroEval Italian Summarization - Ilpost SUM | 39.2 | chrF3++ (%) | 100 |
| MT-Bench PL - Coding | 8.3 | Judge Score (0-10) | 100 |
| VLM-DeflectionBench - Parametric Hallucination Rate | 26.6 | Hallucination rate (%): share of answers that are wrong rath | 100 |
| EuroEval Catalan Summarization - Dacsa CA | 39.98 | chrF3++ (%) | 99.4 |
| EuroEval Dutch Summarization - Wiki Lingua NL | 36.35 | chrF3++ (%) | 99.3 |
| EuroEval Ukrainian Summarization - LR SUM UK | 32.26 | chrF3++ (%) | 99.3 |
| EuroEval Italian Common Sense Reasoning | 84.87 | Common Sense Reasoning Average Score (%) | 99.2 |
| EuroEval Albanian NLU - MMS SQ | 30.53 | Matthews correlation coefficient (%) | 99 |
| EuroEval German Common Sense Reasoning | 84.1 | Common Sense Reasoning Average Score (%) | 99 |
| EuroEval English Common Sense Reasoning | 90.82 | Common Sense Reasoning Average Score (%) | 98.7 |
| EuroEval Icelandic NLU - NQII | 63.53 | Character F1 (%) | 98.7 |
| EuroEval Icelandic Summarization - RRN | 39.67 | chrF3++ (%) | 98.7 |
Interactive version: theaggregate.ai/model?slug=mistral-small-3-1 · How It Works · Data refreshed daily, snapshot 2026-10-20.