A college-level multimodal benchmark that evaluates cross-disciplinary reasoning and the application of expert knowledge to text-and-image problems.
Browse the latest scores, model modes, release dates, and parameter sizes for MMMU.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
| Rank | Model | License | |||
|---|---|---|---|---|---|
 GPT-5.2 Thinking Level · Extra High | 85.90 | 2025-12-11 | Unknown | Closed | |
GPT-5.1 Thinking Level · High | 85.40 | 2025-11-12 | Unknown | Closed | |
GPT-5.1 Thinking Level · High | 85.40 | 2025-11-12 | Unknown | Closed | |
4 |  Doubao Seed 2.0 Pro Thinking Enabled | 85.40 | 2026-02 | Unknown | Closed |
5 |  Qwen3.5-397B-A17B Thinking Enabled | 85.00 | 2026-02-16 | 39.7B | Free Commercial |
6 | GPT-5 Thinking Level · High | 84.20 | 2025-08-07 | Unknown | Closed |
7 | OpenAI o3 Standard Mode | 82.90 | 2025-04-16 | Unknown | Closed |
8 | OpenAI o3 Thinking Enabled | 82.90 | 2025-04-16 | Unknown | Closed |
9 | Qwen3.5-27B Thinking Enabled | 82.30 | 2026-02-25 | 27B | Free Commercial |
10 |  Gemini 2.5-Pro Thinking Enabled | 82.00 | 2025-06-05 | Unknown | Closed |
11 |  Opus 4.5 Extended Thinking | 80.70 | 2025-11-25 | Unknown | Closed |
12 | Gemini 3.1 Pro Preview Thinking Level · High | 80.50 | 2026-02-20 | Unknown | Closed |
13 | GPT-5.2 Thinking Level · Extra HighTools | 80.40 | 2025-12-11 | Unknown | Closed |
14 | Gemini-2.5-Pro-Preview-05-06 Thinking Enabled | 79.60 | 2025-05-06 | Unknown | Closed |
15 | Claude Sonnet 4.5 Thinking Enabled | 77.80 | 2025-09-30 | Unknown | Closed |
16 | Claude Opus 4.6 Extended ThinkingTools | 77.30 | 2026-02-05 | Unknown | Closed |
17 | Claude Sonnet 4 Standard Mode | 76.50 | 2025-05-23 | Unknown | Closed |
18 |  Step3 Standard Mode | 74.20 | 2025-07-31 | 321B | Free Commercial |
19 | Claude Opus 4.6 Extended Thinking | 73.90 | 2026-02-05 | Unknown | Closed |
20 | Haiku 4.5 128K | 73.20 | 2025-10-15 | Unknown | Closed |
21 | Gemini 2.5 Flash-Lite-Preview-09-2025 Standard Mode | 72.70 | 2025-09-25 | Unknown | Closed |
22 | Qwen2.5-VL-72B-Instruct Standard Mode | 70.20 | 2025-01-28 | 72B | Free Commercial |
23 |  ERNIE-4.5-VL-424B-A47B-Base Thinking Enabled | 70.00 | 2025-06-30 | 424B | Free Commercial |
24 | Qwen3-VL-8B-Instruct Standard Mode | 69.60 | 2025-10-15 | 8.8B | Free Commercial |
25 | GPT-5.4 nano Thinking Level · Extra HighTools | 69.50 | 2026-03-17 | Unknown | Closed |
26 | Qwen3-VL-4B-Instruct Standard Mode | 67.40 | 2025-10-15 | 4B | Free Commercial |
27 | GPT-5.4 nano Thinking Level · Extra High | 66.10 | 2026-03-17 | Unknown | Closed |
28 | Gemma 3 - 27B (IT) Standard Mode | 64.90 | 2025-03-12 | 27B | Free Commercial |
29 | GPT-5-Nano Standard Mode | 57.60 | 2025-08-07 | Unknown | Closed |
