← Benchmarks
OCRBench_V2
OCRBench v2: Enhanced large-scale bilingual benchmark for evaluating Large Multimodal Models on visual text localization and reasoning with 10,000 human-verified question-answering pairs across 8 core OCR capabilities
id ocrbench-v2 · max 1 · 8 models reported
| # | Model | Score |
|---|
No scores for this benchmark yet.