OpenCompass Multimodal - MMBench: leaderboard
OpenCompass multimodal score on MMBench visual-language understanding tasks.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Estimated already saturated. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 88.3 | #152 |
| 2 | InternVL3-78B | 87.7 | #352 |
| 3 | InternVL3-38B | 86.8 | #405 |
| 4 | GPT-5 | 86.6 | #98 |
| 5 | GPT-4.1 (2025-04-14) | 86.6 | #250 |
| 6 | GPT-5 Mini (2025-08-07) | 86.2 | #172 |
| 7 | GPT-4o ChatGPT | 86 | |
| 8 | Gemini 2.0 Flash | 85.2 | #342 |
| 9 | InternVL3-14B | 83.6 | #505 |
| 10 | GPT-4.5 | 83.4 | #274 |
| 11 | Grok 2 (1212) | 82.9 | #496 |
| 12 | Qwen 2.5 VL 7B | 82.2 | #643 |
| 13 | InternVL3-8B | 82.1 | #620 |
| 14 | Doubao-1.5-Pro | 82.1 | #287 |
| 15 | GPT-4.1 Mini | 80.9 | #355 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-multimodal-mmbench · How It Works · Data refreshed daily, snapshot 2026-10-20.