OpenCompass Multimodal - MMBench: leaderboard
Metric: Score (%). Source: rank.opencompass.org.cn. 105 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 88.3 |
| 2 | InternVL3-78B | 87.7 |
| 3 | InternVL3-38B | 86.8 |
| 4 | GPT-5 | 86.6 |
| 5 | GPT-4.1 (2025-04-14) | 86.6 |
| 6 | GPT-5 Mini (2025-08-07) | 86.2 |
| 7 | GPT-4o ChatGPT | 86 |
| 8 | Gemini 2.0 Flash | 85.2 |
| 9 | GPT-4.5 | 83.4 |
| 10 | Grok 2 (1212) | 82.9 |
| 11 | InternVL3-8B | 82.1 |
| 12 | Doubao-1.5-Pro | 82.1 |
| 13 | GPT-4.1 Mini | 80.9 |
| 14 | Claude 3.7 Sonnet | 79.7 |
| 15 | GPT-5 Nano | 79.2 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-multimodal-mmbench · How It Works · Data refreshed daily, snapshot 2026-09-05.