Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

MMMU

A benchmark designed to evaluate multimodal models on massive multi-discipline tasks demanding college-level subject knowledge and deliberate reasoning. Includes 11.5K multimodal questions from college exams, quizzes, and textbooks, covering six core disciplines.
Source:

Model Performance

#1
Gemini-2.5-Pro-0325
81.3%
#2
Gemini-2.5-Pro-Thinking
81.3%
#3
OpenAI-O3-medium
80.4%
#4
OpenAI-O3-high
80.4%
#5
OpenAI-O4-mini-medium
79.7%
#6
OpenAI-O4-mini-high
79.7%
#7
GPT-5-mini
78.9%
#8
Claude-4.1-Opus-Thinking
77.5%
#9
OpenAI-O1-1217
77.4%
#10
Grok-4
76.3%
#11
Grok-4-Thinking
76.3%
#12
Claude-3.7-Sonnet-Thinking
75.1%
#13
Claude-4.0-Sonnet-Thinking
74.9%
#14
Claude-4.1-Opus
74.0%
#15
Claude-4.0-Opus
73.3%
#16
Claude-4.0-Sonnet
72.4%
#17
GPT-4.1
72.4%
#18
Llama-4-Maverick-17B
71.7%
#19
Claude-3.7-Sonnet
71.5%
#20
Gemini-2.0-Flash
69.8%
#21
Claude-3.5-Sonnet-1022
68.8%
#22
Gemini-1.5-Pro
65.5%
#23
Mistral-Medium-3.1
63.0%
#24
GPT-4-mini
56.6%