OpenCompass Multimodal - Average: leaderboard
The overall average of the OpenCompass multimodal leaderboard — Shanghai AI Lab's vision-language benchmark aggregate across its component benchmarks.
Metric: Average Score. Source: rank.opencompass.org.cn. Saturation forecast: Around January 2027. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 80.1 | #152 |
| 2 | GPT-5 | 79.9 | #98 |
| 3 | InternVL3-78B | 79.1 | #352 |
| 4 | GPT-5 Mini (2025-08-07) | 78 | #172 |
| 5 | InternVL3-38B | 77.8 | #405 |
| 6 | Doubao-1.5-Pro | 76.5 | #287 |
| 7 | GPT-4.1 (2025-04-14) | 75.9 | #250 |
| 8 | GPT-4o ChatGPT | 75.4 | |
| 9 | GPT-4.5 | 75.3 | #274 |
| 10 | InternVL3-14B | 75.2 | #505 |
| 11 | Gemini 2.0 Flash | 74.3 | #342 |
| 12 | InternVL3-8B | 73.6 | #620 |
| 13 | GPT-5 Nano | 72.2 | #425 |
| 14 | Qwen 2.5 VL 7B | 70.9 | #643 |
| 15 | Claude 3.7 Sonnet | 70.1 | #249 |
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-average · How It Works · Data refreshed daily, snapshot 2026-10-20.