Mistral Instruct v0.3 7B — benchmark results
Provider: Mistral. Released 2024-05-22. Access: Open.
Unified ELO 1231 ± 64, rank #1689 of 1839 rated models, from 7 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Safety XSTest | 92.6 | LM Evaluated Safety score (%) | 19.2 |
| HELM Safety BBQ | 66.8 | BBQ accuracy (%) | 7 |
| HELM Safety SimpleSafetyTests | 81 | LM Evaluated Safety score (%) | 5.2 |
| HELM Safety | 73 | Mean score (self-reported) | 5.1 |
| HELM Safety Anthropic Red Team | 90.1 | LM Evaluated Safety score (%) | 4.7 |
| HELM AIR-Bench | 32.6 | Refusal Rate (%) | 3.5 |
| HELM Safety HarmBench | 34.4 | LM Evaluated Safety score (%) | 2.3 |
Interactive version: theaggregate.ai/model?slug=mistral-instruct-v0-3-7b · How It Works · Data refreshed daily, snapshot 2026-08-05.