OCRBench_V2

Multilingual: no

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

Methodology

Imported from llm-stats public benchmark metadata. Modality: multimodal. Max score: 1. Categories: image_to_text, vision. Language: en. Verified by llm-stats: no.

Leaderboard

Showing 4 of 4 results

Nova 2 Pro self-reported llm-stats

64.5%
Nova 2 Omni self-reported llm-stats

58.2%
Qwen2.5-Omni-7B self-reported llm-stats

57.8%
Nova 2 Lite self-reported llm-stats

56.1%