Mixtral-8x22B-v0.1: benchmark results
Mistral's base (non-instruct) sparse mixture-of-experts with 141B total and 39B active parameters and a 64K context, under Apache 2.0 (April 2024). Provider: Mistral. Released 2024-04-10. Access: Open.
Unified ELO 1527 ± 1, rank #557 of 1392 rated models, from 19 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open LLM Leaderboard - GPQA | 16.78 | Score | 94 |
| RABBITS B4B | 78.82 | Accuracy (%) | 88.9 |
| Open LLM Leaderboard - BBH | 45.59 | Score | 85.7 |
| Open LLM Leaderboard - MMLU-Pro | 40.44 | Score | 85.3 |
| RABBITS B4BQA | 98.66 | Accuracy (%) | 82.1 |
| Open PL LLM - Generative | 65.88 | Average Generative Score (%) | 80.8 |
| BenchBench | 73.82 | Aggregate Score (%) | 78.7 |
| Open PL LLM Leaderboard | 60.75 | Average Score (%) | 76.6 |
| Open PL LLM - Multiple Choice | 54.07 | Average Multiple-Choice Score (%) | 74.8 |
| MMLU | 77.8 | Accuracy (%) | 74.4 |
| Open PL LLM - RAG | 68.75 | Average RAG Score (%) | 74.1 |
| Open LLM Leaderboard - MATH Level 5 | 18.35 | Score | 68.1 |
Interactive version: theaggregate.ai/model?slug=mixtral-8x22b-v0-1 · How It Works · Data refreshed daily, snapshot 2026-09-05.