Verified AI benchmark

MMMU Leaderboard

MMMU — official/provider-reported evaluation result. Methodology conditions are preserved in the result notes.

Semantic classTechnical Performance CategoryMultimodal DirectionHigher is better Verified entities15

How to interpret this leaderboard

What this benchmark does—and does not—measure

This is a technical-performance benchmark. Compare scores only within this benchmark definition and version; AI Orbit does not combine it with product-experience or otherwise incompatible semantic classes.

Method & source

Unit: %
Results: 15
Latest test: Aug 7, 2025
Distinct source links: 5

Open official methodology / source

Leading score84.20 %

Entities at top score1

Evidence freshnessAug 7, 2025

Verified results

MMMU rankings

Only the latest verified result for each listed entity is shown. Equal scores receive the same displayed rank.

Rank	Model / Tool	Score	Tested	Evidence
#1	GPT-5	84.20%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#2	o3	82.90%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#3	Gemini 2.5 Pro	82.00%	Not recorded	Google DeepMind
#4	o4-mini	81.60%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#4	GPT-5 mini	81.60%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#6	Gemini 2.5 Flash	79.70%	Not recorded	Google DeepMind
#7	Claude Opus 4	76.50%	May 22, 2025	Anthropic Claude 4 announcement
#8	GPT-5 nano	75.60%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#9	Claude 3.7 Sonnet	75.00%	May 22, 2025	Anthropic Claude 4 announcement
#10	GPT-4.1	74.80%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#11	Claude Sonnet 4	74.40%	May 22, 2025	Anthropic Claude 4 announcement
#12	GPT-4.1 mini	72.70%	Aug 7, 2025	OpenAI GPT-5 developer announcement
#13	GPT-4o	68.70%	Apr 14, 2025	OpenAI GPT-4.1 announcement
#14	GPT-4o mini	56.30%	Apr 14, 2025	OpenAI GPT-4.1 announcement
#15	GPT-4.1 nano	55.40%	Aug 7, 2025	OpenAI GPT-5 developer announcement

Limitations

What to keep in mind

A benchmark is one evidence signal, not a universal product verdict. Test conditions, benchmark versions, source methodology and date can all affect interpretation. Compare shared benchmarks and current product evidence before making a model or tool decision.

Freshness

Why dates and sources are shown

AI Orbit retains the source and tested date for each verified result so readers can judge whether evidence is current enough for their use case. Older evidence is not silently presented as a new test.