Benchmark

MMMU Benchmark

What is MMMU?

Image preprocessing and chain-of-thought settings affect results. Benchmark Radar tracks 4 reported scores for MMMU, cited by 10 source documents.

Released: 2023-11-27
Evaluates: multimodal
Openness: unknown
Importer: Model reports
Reported scores: 4

Source provenance

Original evidence https://mmmu-benchmark.github.io/

MMMU links

Website https://mmmu-benchmark.github.io/

MMMU results and reported scores

Model reports

Model	Organization	Reported value	Reported	Evidence
Claude Opus 4	Anthropic	76.5	2025-05-22	evidence
Claude Sonnet 4	Anthropic	74.4	2025-05-22	evidence
Gemini 2.5 Pro	Google	82.0	2025-06-17	evidence
Qwen3.5-397B-A17B	Qwen	85.0	2026-02-16	evidence

Scores are partitioned by the source that reported them and are never merged into a single cross-source ranking, because the sources measure different things and say so.

Which sources cite MMMU?

Qwen3.5-397B-A17B Qwen, 2026-02-16
GPT-5 OpenAI, 2025-08-07
Grok 4 xAI, 2025-07-09
Gemini 2.5 Pro Google, 2025-06-17
Claude Opus 4 and Sonnet 4 Anthropic, 2025-05-22
o3 and o4-mini OpenAI, 2025-04-16
GPT-4.1 OpenAI, 2025-04-14
Llama 4 Meta, 2025-04-05
Claude 3.7 Sonnet Anthropic, 2025-02-24
Gemini 1.5 Google, 2024-03-08