OpenCompass Multimodal - OCRBench: leaderboard
OpenCompass multimodal score on OCRBench text recognition and document understanding tasks.
Metric: Score (normalized). Source: rank.opencompass.org.cn. 105 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | JT-VL-Chat-2B | 94.5 | |
| 2 | JT-VL-Chat-Mini | 94.2 | |
| 3 | CongRong-v2.0 | 92.7 | |
| 4 | Step-1o | 92.6 | |
| 5 | JT3.5 | 91.5 | |
| 6 | SAIL-VL2-8B | 91.2 | |
| 7 | MUG-U-7B | 91.1 | |
| 8 | Ovis2-4B | 91.1 | |
| 9 | TeleMM-2.0-Thinking | 90.8 | |
| 10 | InternVL3-78B | 90.8 | #352 |
Interactive version: theaggregate.ai/benchmark?slug=opencompass-multimodal-ocrbench · How It Works · Data refreshed daily, snapshot 2026-10-20.