MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark)
A benchmark of ~11.5K college-level multimodal questions spanning 30 subjects and 183 subfields across six disciplines, measuring a vision-language model's accuracy at jointly perceiving images (charts, diagrams, maps, tables, etc.) and reasoning with domain knowledge.
What this benchmark measures
A benchmark of ~11.5K college-level multimodal questions spanning 30 subjects and 183 subfields across six disciplines, measuring a vision-language model's accuracy at jointly perceiving images (charts, diagrams, maps, tables, etc.) and reasoning with domain knowledge.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is accuracy. It should be interpreted within MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark), not compared as part of a site-wide ranking.
Frequently asked
What is MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark)?
A benchmark of ~11.5K college-level multimodal questions spanning 30 subjects and 183 subfields across six disciplines, measuring a vision-language model's accuracy at jointly perceiving images (charts, diagrams, maps, tables, etc.) and reasoning with domain knowledge. It is a multimodal benchmark measured by accuracy.
What does accuracy mean on MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark)?
MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark) reports accuracy (%); higher is better. Scores are shown only within MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark) and are never averaged with other benchmarks.
What is the top reported MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark) score?
GPT-5.1 has the top reported score on MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark): 85.4% (accuracy).
Why do MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark) scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark) scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".