Meta Llama 3 70B Instruct — Benchmarks

Benchmark scores for Meta Llama 3 70B Instruct aggregated from public leaderboards, with how it ranks among open models. See hardware requirements for what you need to run it.

Overall rank: #58 of 73 open modelscomposite 32/100 across 5 benchmarks in 3 categories · methodology

Knowledge

BenchmarkScoreOpen rankAll models
MMLU-Pro56.2%#57 / 119#153 / 259
MMLU79.3%#12 / 76#29 / 136

Math

BenchmarkScoreOpen rankAll models
AIME 2024/20254.3%#26 / 34#138 / 155
MATH Level 522.6%#25 / 32#88 / 108

Reasoning

BenchmarkScoreOpen rankAll models
GPQA Diamond40.6%#32 / 46#147 / 182

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