Mistral 7B v0.1 — Benchmarks

Benchmark scores for Mistral 7B v0.1 aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.

Overall rank: #28 of 123 open modelscomposite 59.6/100 across 11 benchmarks in 3 categories · methodology

Knowledge

BenchmarkScoreOpen rankAll models
MMLU62.5%#40 / 77#86 / 136
HellaSwag81.0%#13 / 42#26 / 76
ARC Challenge78.6%#12 / 51#18 / 77
BoolQ87.4%#5 / 33#15 / 77
OpenBookQA79.8%#5 / 21#7 / 42
TriviaQA75.2%#8 / 18#22 / 39
MMLU-Pro30.9%#95 / 119#225 / 259

Math

BenchmarkScoreOpen rankAll models
GSM8K54.4%#27 / 60#40 / 93

Reasoning

BenchmarkScoreOpen rankAll models
BIG-Bench Hard56.1%#13 / 37#20 / 50
WinoGrande75.3%#17 / 46#36 / 80
PIQA83.0%#8 / 35#14 / 60

Scores aggregated from public benchmark sources (each linked from the benchmark pages). llmrun does not run these benchmarks.