OCRBench v2: leaderboard
OCRBench v2 evaluates large multimodal models on bilingual visual text localization and reasoning tasks.
Metric: Average (self-reported). Source: benchmarklist.com. Status: saturation imminent. 31 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 9B | 64.1 |
| 2 | Gemini 3 Pro (Preview) | 63.8 |
| 3 | Gemini 2.5 Pro | 62.2 |
| 4 | Qwen3 Omni 30B A3B Instruct | 61.3 |
| 5 | Claude Opus 4.6 | 59.8 |
| 6 | GPT-5.2 | 52.6 |
| 7 | Molmo2-8B | 44.7 |
| 8 | Ministral 3 14B | 44 |
| 9 | Phi-4 Multimodal Instruct | 38.1 |
Interactive version: theaggregate.ai/benchmark?slug=ocrbench-v2 · How It Works · Data refreshed daily, snapshot 2026-09-05.