Multimodal · benchmark

MMMU

Test expert-level reasoning across images and academic disciplines.

What it measures

Validation / test accuracy (%)

Validation, test and Pro results must not be combined.

Public evaluation code. Check upstream code and dataset terms.

Open source / dataset · Run instructions · Official results

Reported ranking tracks

• MMMU · validation · 189 entries · captured 2026-10-07T07:01:28Z

Explore all evaluations · Find a competition