MMMU-Pro — MMMU-Pro w/ python tool enabled; tools axis distinct from base.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | Gemini 3.1 Pro | 85.3 | 2 | 2026-05-03 | |
| 2 | GPT-5.4 | OpenAI | 82.1 | 1 | 2026-05-03 |
| 3 | Kimi K2.6 | Moonshot | 80.1 | 1 | 2026-05-03 |
| 4 | Kimi K2.5 | Moonshot | 77.7 | 1 | 2026-05-03 |
| 5 | Claude Opus 4.6 | Anthropic | 77.3 | 1 | 2026-05-03 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.