Mixtral 8x7B: benchmark results
Mistral's first Apache-2.0 sparse-MoE model (46.7B total/12.9B active, 8 experts pick 2), December 2023; this row is the pretrained base, non-instruct weights. Provider: Mistral. Released 2023-12-11. Access: Open.
Aggregate Rating 1421 ± 13, rank #1143 of 1656 rated models, from 60 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BizBench - FormulaEval | 94 | Accuracy (%; 50 financial-formula functions to complete from | 93.3 |
| XDomainBench | 56.4 | R (k=1, Deterministic) (self-reported) | 92.3 |
| MT-Bench PL - Writing | 9.35 | Judge Score (0-10) | 90.8 |
| BizBench - ConvFinQA Extract | 92.8 | Accuracy (%; 916 ConvFinQA questions answerable by extractin | 90 |
| BizBench - TAT-QA Extract | 91.1 | Accuracy (%; 248 TAT-QA questions answerable by extracting a | 86.7 |
| BizBench | 69.9 | Mean accuracy (%) over the eight BizBench tasks (FinKnow, Fo | 85.7 |
| BizBench - CodeFinQA | 58.5 | Accuracy (%; 844 FinQA questions over report text and tables | 80 |
| BizBench - CodeTAT-QA | 83.9 | Accuracy (%; 392 TAT-QA table questions answered by generati | 80 |
| BizBench - FinKnow | 47.1 | Accuracy (%; 877 multiple-choice business and finance knowle | 80 |
| MT-Bench PL - Roleplay | 8.95 | Judge Score (0-10) | 75.5 |
| BizBench - FinCode | 19 | Accuracy (%; 121 CFA and CPA exam questions answered by gene | 73.3 |
| BizBench - SEC-Num | 73.1 | Accuracy (%; 2,000 SEC 10-K and 10-Q filing snippets, extrac | 73.3 |
Interactive version: theaggregate.ai/model?slug=mixtral-8x7b · How It Works · Data refreshed daily, snapshot 2026-10-20.