General OCR Capabilities
Comprehensive benchmarks covering multiple aspects of OCR performance.
4
Datasets
52
Results
overall-en-private
Canonical metric
Canonical Benchmark
OCRBench v2
Tests 8 core OCR capabilities across 23 tasks. Evaluates LMMs on text recognition, referring, extraction.
Primary metric: overall-en-private
Top 10
Leading models on OCRBench v2.
| Rank | Model | overall-en-private | Year | Source |
|---|---|---|---|---|
| 1 | seed-1.6-vision | 62.2 | 2025 | paper |
| 2 | gemini-25-pro | 62.2 | 2025 | paper |
| 3 | qwen3-omni-30b | 61.3 | 2025 | paper |
| 4 | nemotron-nano-v2-vl | 61.2 | 2025 | paper |
| 5 | Qianfan-OCR | 60.8 | 2026 | paper |
| 6 | gemini-25-pro | 59.3 | 2025 | paper |
| 7 | minicpm-v-4.5-8b | 58.8 | 2025 | paper |
| 8 | sail-vl2-8b | 57.6 | 2025 | paper |
| 9 | llama-3.1-nemotron-nano-vl-8b | 56.4 | 2025 | paper |
| 10 | Qianfan-OCR | 56.0 | 2026 | paper |
All datasets
4 datasets tracked for this task.
Related tasks
Other tasks in Computer Vision.