Mixtral Instruct 8x22B — benchmark results
Provider: Mistral. Released 2024-04-17. Access: Open.
Unified ELO 1429 ± 25, rank #1101 of 1839 rated models, from 29 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Finance - FinQA | 44 | Program Accuracy | 66.7 |
| HELM EWoK - Material Properties | 90.32 | EM | 64.3 |
| HELM Finance - BANKING77 | 73.1 | EM | 61.9 |
| HELM EWoK - Agent Properties | 85.85 | EM | 50 |
| HELM EWoK - Physical Interactions | 74.29 | EM | 50 |
| HELM EWoK - Physical Relations | 75.78 | EM | 50 |
| HELM EWoK - Spatial Relations | 66.33 | EM | 42.9 |
| HELM ThaiExam - TPAT-1 | 42.24 | EM | 39 |
| HELM Safety BBQ | 91.6 | BBQ accuracy (%) | 36 |
| HELM Finance - FinanceBench | 52.14 | Correct Answer | 33.3 |
| HELM ThaiExam - TGAT | 63.08 | EM | 32.9 |
| HELM Safety XSTest | 94.1 | LM Evaluated Safety score (%) | 30.8 |
Interactive version: theaggregate.ai/model?slug=mixtral-instruct-8x22b · How It Works · Data refreshed daily, snapshot 2026-08-05.