MMLU

A 57-subject multiple-choice exam — history, law, medicine, maths — that was the standard measure of how much a model knows from 2020 until roughly 2024, when frontier scores crowded into the high 80s and labs moved on to harder tests. The scores here were published years apart under different testing setups, so read them as a historical record rather than a like-for-like ranking. Higher is better.

gpt-oss-120b by OpenAI holds the best MMLU result among tracked models, at 90% (released Aug 5 2025). Higher scores are better on this benchmark.

1. gpt-oss-120b (OpenAI) — 90%; 2. GPT-4 (OpenAI) — 86.4%; 3. gpt-oss-20b (OpenAI) — 85.3%; 4. Claude 2 (Anthropic) — 78.5%.

4 tracked models have a published MMLU score. Scores are the figures reported by each lab at that model's release, so this page is a record of results over time rather than a re-run leaderboard.

gpt-oss-120b by OpenAI is the highest-ranked model with downloadable weights on MMLU, scoring 90% at rank 1 overall.