88.2%
#1 of 14
MMMU accuracy of Gemini 3.1 Pro Preview in February 2026, the highest of 14 models; the best human expert scored 88.6%
What the chart shows
MMMU tests college-level reasoning on questions that combine text with diagrams, charts, tables and equations. In February 2026 Gemini 3.1 Pro Preview led with 88.2%, 0.4 points below the best human expert reference of 88.6%. Gemini 3 Flash (87.6%), Gemini 3 Pro (87.5%) and GPT 5.2 (86.7%) followed.
The 14 models span 7.8 points, from 88.2% down to o3 at 80.4%. Of the models listed as 2026 releases, Kimi K2.5 scored highest at 84.3%.
Key numbers
| Measure | Value |
|---|---|
| Gemini 3.1 Pro Preview | 88.21% |
| Gemini 3 Flash (12/25) | 87.63% |
| Gemini 3 Pro (11/25) | 87.51% |
| GPT 5.2 | 86.67% |
| o3 | 80.42% |
| Best human expert (report) | 88.6% |
How to read it
- Scores from Vals.ai’s independent MMMU evaluation, February 2026.
- The best human expert reference (88.6%) comes from the report’s chart; it is not in the data file and is drawn as a dashed line.
- The report says the chart shows the top 15 models; the data file has 14.
Source: Vals.ai, 2026, via the AI Index Report 2026 (Stanford HAI), figure 2.4.1: MMMU: accuracy. Chart and text by AI Adoption.
Numbers checked against the published data on 20 October 2026.
Download the data (CSV) · Cite this chart
Numbers checked against the published data on 20 October 2026.
Download the data (CSV) · Cite this chart
Discussion