Mixtral 8x7B (v0.1) — benchmark results

Mistral's Apache-2.0 sparse-MoE model (46.7B total/12.9B active, 32K context) that beat Llama 2 70B on most benchmarks at launch. Provider: Mistral. Released 2023-12-11. Access: Open.

Unified ELO 1454 ± 13, rank #978 of 1776 rated models, from 105 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
EuroEval Portuguese NLU - MultiWikiQA PT77.35Reading comprehension Score (%)94
EuroEval French NLU - FQuAD74.48Reading comprehension Score (%)93.3
European LLM Leaderboard - Zero-Shot Accuracy56.24Average Accuracy (%)92.7
OpenBookQA85.8Accuracy (%)90.5
ARC Challenge (AI2)87.3Accuracy (%)89.7
EuroEval Finnish NLU - Tydiqa FI72.27Reading comprehension Score (%)89.5
HellaSwag86.7Accuracy (%)89.5
URIAL-Bench - Coding5.3Judge Score (0-10)88.9
EuroEval Spanish NLU - MLQA ES66.3Reading comprehension Score (%)87.5
EuroEval Italian NLU - SQuAD IT72.85Reading comprehension Score (%)87.1
EuroEval Danish Knowledge87.29Knowledge Average Score (%)87
SpeechMap Compliance86.7% Requests Completed84.2

Interactive version: theaggregate.ai/model?slug=mixtral-8x7b-v0-1 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.