Which AI model scores highest on MMMU-Pro?

GPT-5 by OpenAI holds the best MMMU-Pro (standard and vision mean) result among tracked models, at 78.4% (released Aug 7 2025). Higher scores are better on this benchmark.

What are the top 5 models on MMMU-Pro?

1. GPT-5 (OpenAI) — 78.4%; 2. GPT-5 mini (OpenAI) — 74.1%; 3. GPT-5 nano (OpenAI) — 62.6%; 4. LLaMA 4 Maverick (Meta) — 59.6%; 5. LLaMA 4 Scout (Meta) — 52.2%.

How many models have a published MMMU-Pro score?

5 tracked models have a published MMMU-Pro (standard and vision mean) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.

Multimodal reasoning

MMMU-Prostandard and vision mean

average across the standard and vision sets

standard and vision mean with tools MMMU-Pro

Rankings

Higher is better

MMMU-Pro — frequently asked questions

: average across the standard and vision sets
: GPT-5 by OpenAI holds the best MMMU-Pro (standard and vision mean) result among tracked models, at 78.4% (released Aug 7 2025). Higher scores are better on this benchmark.
: 1. GPT-5 (OpenAI) — 78.4%; 2. GPT-5 mini (OpenAI) — 74.1%; 3. GPT-5 nano (OpenAI) — 62.6%; 4. LLaMA 4 Maverick (Meta) — 59.6%; 5. LLaMA 4 Scout (Meta) — 52.2%.
: 5 tracked models have a published MMMU-Pro (standard and vision mean) score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.