benchmark evidence
MMMU
Massive Multidiscipline Multimodal Understanding — 11K expert questions across 57 subjects.
activeupstream source ->
winner on MMMU
OpenAI: GPT-5.185.4
direct benchmark result, not a broad vertical composite | source row dated 2026-05-15
scored on 2026-05-15 · stale source data (55d)
latest mapped results | top 20
| # | Model | Score | Evidence | Tested |
|---|---|---|---|---|
| 1 | OpenAI: GPT-5.1 | 85.4 | model-only independent_benchmark | 2026-05-15 |
| 2 | OpenAI: o3 | 82.9 | model-only independent_benchmark | 2000-01-01 |
| 3 | OpenAI: o4 Mini | 81.6 | model-only independent_benchmark | 2000-01-01 |
| 4 | Google: Gemini 2.5 Flash | 79.7 | model-only independent_benchmark | 2000-01-01 |
| 5 | Google: Gemini 2.5 Pro | 79.6 | model-only independent_benchmark | 2000-01-01 |
| 6 | Anthropic: Claude Sonnet 4.5 | 77.8 | model-only independent_benchmark | 2026-05-15 |
| 7 | OpenAI: o1 | 77.6 | model-only independent_benchmark | 2000-01-01 |
| 8 | Anthropic: Claude Opus 4.7 | 76.5 | model-only independent_benchmark | 2026-05-08 |
| 9 | Anthropic: Claude Opus 4.5 | 76.5 | model-only independent_benchmark | 2026-05-15 |
| 10 | OpenAI: GPT-4.1 | 74.8 | model-only independent_benchmark | 2000-01-01 |
| 11 | Anthropic: Claude Sonnet 4 | 74.4 | model-only independent_benchmark | 2026-05-15 |
| 12 | Anthropic: Claude Sonnet 4.6 | 74.4 | model-only independent_benchmark | 2026-05-08 |
| 13 | Meta: Llama 4 Maverick | 73.4 | model-only independent_benchmark | 2026-05-15 |
| 14 | Google: Gemini 2.0 Flash Lite | 71.7 | model-only independent_benchmark | 2026-05-15 |
| 15 | Google: Gemini 2.0 Flash | 70.7 | model-only independent_benchmark | 2000-01-01 |
| 16 | Meta: Llama 4 Scout | 69.4 | model-only independent_benchmark | 2026-05-15 |
| 17 | Google: Gemma 3 27B | 64.9 | model-only independent_benchmark | 2026-05-15 |
| 18 | Amazon: Nova Pro 1.0 | 62.0 | model-only independent_benchmark | 2026-05-15 |
| 19 | OpenAI: GPT-5 | 56.8 | model-only independent_benchmark | 2026-05-15 |
what this result means
Massive Multidiscipline Multimodal Understanding — 11K expert questions across 57 subjects.
This benchmark contributes direct public evidence. Read its scope before generalizing the result.
A win here is a win on MMMU. Broad task pages require independent corroboration before naming a general winner.
source record
category: reasoning
metric: accuracy
matched models: 19
latest source date: 2026-05-15
direction: higher is better