OCRBench-V2 (zh): leaderboard
OCRBench v2 Chinese subset: Enhanced benchmark for evaluating Large Multimodal Models on visual text localization and reasoning with Chinese text content.
Metric: Score (self-reported). Source: benchmarklist.com. Status: saturation imminent. 26 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.7 Plus | 67.1 |
| 2 | Qwen 3.5 9B | 64.1 |
| 3 | Gemini 3 Pro (Preview) | 63.8 |
| 4 | Qwen 3.6 Plus | 63.6 |
| 5 | Gemini 2.5 Pro | 62.2 |
| 6 | Qwen3 Omni 30B A3B Instruct | 60 |
| 7 | Gemini 3.1 Pro (Preview) | 58.2 |
| 8 | GPT-5.4 (xHigh) | 57.7 |
| 9 | Claude Opus 4.6 (Max) | 54.9 |
| 10 | GPT-5.2 | 52.6 |
| 11 | Ministral 3 14B | 44 |
| 12 | Phi-4 Multimodal Instruct | 37.3 |
| 13 | Molmo2-8B | 21.1 |
Interactive version: theaggregate.ai/benchmark?slug=ocrbench-v2-zh · How It Works · Data refreshed daily, snapshot 2026-09-05.