Mistral Small: benchmark results
Mistral's small-tier workhorse under a rolling name, spanning the 22B (2409, research license) and 24B Apache-2.0 (2501) releases. Provider: Mistral. Released 2025-01-30. Access: Open.
Aggregate Rating 1514 ± 17, rank #740 of 1656 rated models, from 27 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| InfiBench | 55.62 | Score (%) | 87.6 |
| ContextEcho | 27 | Delta (self-reported) | 81.8 |
| SWE-PRBench - Hallucination Rate | 25.1 | Hallucination rate (%): share of the model's review comments | 71.4 |
| ProLLM - Entity Extraction | 81.8 | Score (%) | 58.5 |
| ProLLM - Function Calling | 87 | Score (%) | 52.8 |
| MixEval | 46.2 | Score | 52 |
| SWE-PRBench (Config A) - Detection Rate | 25.7 | Detection rate (%): share of the human-flagged issues that t | 42.9 |
| VAmoS Bench | 59.7 | Task Completion (%) | 33.3 |
| GRIPS | 65.5 | Accuracy (%) | 30.8 |
| Wolfram LLM Benchmarking Project | 29.9 | Correct Functionality (%) | 30.2 |
| SWE-PRBench (Config A) | 13.1 | Composite review score (0-1, scaled to 0-100): per pull requ | 28.6 |
| SWE-PRBench (Config C) | 9.1 | Composite review score (0-1, scaled to 0-100): per pull requ | 28.6 |
Interactive version: theaggregate.ai/model?slug=mistral-small · How It Works · Data refreshed daily, snapshot 2026-10-20.