Mixtral 8x22B Instruct (v0.1): benchmark results
Mistral Mixtral 8x22B Instruct v0.1 checkpoint. Provider: Mistral. Released 2024-04-17. Access: Open.
Aggregate Rating 1509 ± 9, rank #756 of 1656 rated models, from 352 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ClinicalMC (English) | 49.8 | Mean of 11 task scores (%; triage accuracy, examination reca | 100 |
| OpenCompass LLM - Knowledge - English (CompassBench 2405) | 83.7 | Score (%) | 100 |
| OpenCompass Math - Primary School - English (CompassBench 2405) | 90 | Score (%) | 100 |
| Open LLM Leaderboard v1 - GSM8K | 82.03 | Accuracy (%) (5-shot) | 99.8 |
| Open LLM Leaderboard v1 - MMLU | 77.77 | Accuracy (%) (5-shot) | 98.9 |
| MERA - RCB | 61.19 | Accuracy (%) | 98.3 |
| Open LLM Leaderboard v1 - WinoGrande | 85.16 | Accuracy (%) (5-shot) | 97.5 |
| Open PL LLM - CBD (multiple choice, 5-shot) | 41.23 | Macro F1 (%) | 97.2 |
| Open PL LLM - DYK (multiple choice, 5-shot) | 70.93 | Binary F1 (%) | 96.9 |
| Open LLM Leaderboard v1 - HellaSwag | 89.08 | Normalized accuracy (%) (10-shot) | 96.8 |
| Open PL LLM - PSC (generative, 5-shot) | 98.47 | Binary F1 (%) | 96.2 |
| Open PL LLM - Generative | 68.64 | Average Generative Score (%) | 94.8 |
Interactive version: theaggregate.ai/model?slug=mixtral-8x22b-instruct-v0-1 · How It Works · Data refreshed daily, snapshot 2026-10-20.