MMMU
multimodal
+
+
+
+
About
MMMU (Massive Multi-discipline Multimodal Understanding) is a comprehensive benchmark featuring 11.5K multimodal questions from college exams, quizzes, and textbooks across six core disciplines: Art & Design, Business, Science, Health & Medicine, Humanities & Social Science, and Tech & Engineering. It evaluates expert-level knowledge and reasoning with 30 heterogeneous image types, challenging models to perform tasks requiring college-level subject knowledge and deliberate reasoning.
+
+
+
+
Evaluation Stats
Total Models52
Organizations11
Verified Results0
Self-Reported50
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
52 models
Top Score
84.2%
Average Score
64.1%
High Performers (80%+)
4Top Organizations
#1Anthropic
3 models
72.6%
#2Moonshot AI
1 model
70.0%
#3Google
11 models
68.9%
#4OpenAI
11 models
66.9%
#5Alibaba Cloud / Qwen Team
5 models
65.7%
+
+
+
+
Leaderboard
52 models ranked by performance on MMMU
| License | Links | ||||
|---|---|---|---|---|---|
| Aug 7, 2025 | Proprietary | 84.2% | |||
| Apr 16, 2025 | Proprietary | 82.9% | |||
| Jun 5, 2025 | Proprietary | 82.0% | |||
| Apr 16, 2025 | Proprietary | 81.6% | |||
| May 20, 2025 | Proprietary | 79.7% | |||
| May 20, 2025 | Proprietary | 79.6% | |||
| Feb 17, 2025 | Proprietary | 78.0% | |||
| Dec 17, 2024 | Proprietary | 77.6% | |||
| Jan 21, 2025 | Proprietary | 75.4% | |||
| Feb 27, 2025 | Proprietary | 75.2% |
Showing 1 to 10 of 52 models
...