AI model rankings
Understand capability differences before choosing AI resources.
MMMU · Visual understanding
Historical benchmark · Source not updated for over 60 days; not a current model ranking
MMMU-Pro · Overall · 版本未披露 / 待核验 · Scores range from 0–100; higher is better. Missing scores are excluded, and ties share a rank. Search preserves original ranks.
| Rank | Model and version | Score | Evidence type |
|---|---|---|---|
| 1 | Chance Vision 1.5榜单原始模型版本 · Model date 2026-07-01Model details ↗ | 86.9 | Author-reported score |
| 2 | GPT-5.4 Thinking w/ tools榜单原始模型版本 · Model date 2026-03-05Model details ↗ | 82.1 | Author-reported score |
| 3 | GPT-5.4 Thinking w/o tools榜单原始模型版本 · Model date 2026-03-05Model details ↗ | 81.2 | Author-reported score |
| 4 | Gemini 3.0 Pro榜单原始模型版本 · Model date 2025-11-18Model details ↗ | 81.0 | Author-reported score |
| 5 | Gemini 3.1 Pro Thinking (High)榜单原始模型版本 · Model date 2026-02-19Model details ↗ | 80.5 | Author-reported score |
| 6 | GPT-5.2 Thinking w/o Python榜单原始模型版本 · Model date 2025-12-11Model details ↗ | 80.4 | Author-reported score |
| 6 | Muse Spark Thinking榜单原始模型版本 · Model date 2026-04-08Model details ↗ | 80.4 | Author-reported score |
| 8 | GPT-5.2 Thinking w/o tools榜单原始模型版本 · Model date 2025-12-11Model details ↗ | 79.5 | Author-reported score |
| 9 | GPT-5.1 Thinking榜单原始模型版本 · Model date 2025-11-13Model details ↗ | 79.0 | Author-reported score |
| 10 | GPT-5 w/ thinking榜单原始模型版本 · Model date 2025-08-07Model details ↗ | 78.4 | Author-reported score |
| 11 | Claude Opus 4.6 w/ tools榜单原始模型版本 · Model date 2026-02-05Model details ↗ | 77.3 | Author-reported score |
| 12 | Gemma 4 31B榜单原始模型版本 · Model date 2026-04-02Model details ↗ | 76.9 | Author-reported score |
| 13 | o3榜单原始模型版本 · Model date 2025-04-16Model details ↗ | 76.4 | Author-reported score |
| 14 | GPT-5.1榜单原始模型版本 · Model date 2025-11-13Model details ↗ | 76.0 | Author-reported score |
| 15 | Claude Sonnet 4.6 w/ tools榜单原始模型版本 · Model date 2026-02-17Model details ↗ | 75.6 | Author-reported score |
| 16 | Claude Sonnet 4.6 w/o tools榜单原始模型版本 · Model date 2026-02-17Model details ↗ | 74.5 | Author-reported score |
| 17 | Claude Opus 4.5榜单原始模型版本 · Model date 2025-11-24Model details ↗ | 73.9 | Author-reported score |
| 17 | Claude Opus 4.6 w/o tools榜单原始模型版本 · Model date 2026-02-05Model details ↗ | 73.9 | Author-reported score |
| 19 | Gemma 4 26B A4B榜单原始模型版本 · Model date 2026-04-02Model details ↗ | 73.8 | Author-reported score |
| 20 | dots.vlm1榜单原始模型版本 · Model date 2025-08-05Model details ↗ | 70.1 | Author-reported score |
| 21 | Claude Sonnet 4.5榜单原始模型版本 · Model date 2025-09-29Model details ↗ | 68.9 | Author-reported score |
| 22 | Qwen3-VL 235B-A22B榜单原始模型版本 · Model date 2025-09-22Model details ↗ | 68.1 | Author-reported score |
| 23 | Gemini 2.5 Pro 05-06榜单原始模型版本 · Model date 2025-05-06Model details ↗ | 68.0 | Author-reported score |
| 24 | Seed 1.5-VL Thinking榜单原始模型版本 · Model date 2025-05-13Model details ↗ | 67.6 | Author-reported score |
| 25 | Seed 1.6-Thinking榜单原始模型版本 · Model date 2025-06-25Model details ↗ | 66.4 | Author-reported score |
Source evidence and evaluation methodology
Source:MMMU · Visual understanding · CC BY-SA 4.0. We organize original scores without creating cross-benchmark composite scores. Listing a result does not mean we independently reproduced it.
Attribution: MMMU Benchmark. We adapt the original leaderboard into dimension-specific tables. Adapted data is provided under CC BY-SA 4.0. The original authors do not endorse this site. License and attribution for this source version
Source snapshot:35faf66edfd44373a0b6540e7afe0ede · Source commit：9092d1bda95bd2f957c323fc8957c89db6944c5a · Original content SHA-256:2ec354bc9eab734c1ba6a78f604d2ad7e5581a30257f50ead26523ab630978fb
Fetched on demand and published after admin review, not in real time. Undisclosed evaluation versions are not inferred.Different dimensions and test settings cannot be compared directly.