Wikibench

OCRBench-V2 (zh)

vision

OCRBench v2 Chinese subset: Enhanced benchmark for evaluating Large Multimodal Models on visual text localization and reasoning with Chinese text content

Leaderboard

Showing 11 of 11 results

  1. Qwen3 VL 235B A22B Thinking
    63.5%
    i
  2. Qwen3 VL 32B Thinking
    62.1%
    i
  3. Qwen3 VL 235B A22B Instruct
    61.8%
    i
  4. Qwen3 VL 8B Instruct
    61.2%
    i
  5. Qwen3 VL 30B A3B Thinking
    60.4%
    i
  6. Qwen3 VL 32B Instruct
    59.2%
    i
  7. Qwen3 VL 8B Thinking
    59.2%
    i
  8. Qwen2.5 VL 32B Instruct
    59.1%
    i
  9. Qwen3 VL 30B A3B Instruct
    57.8%
    i
  10. Qwen3 VL 4B Instruct
    57.6%
    i
  11. Qwen3 VL 4B Thinking
    55.8%
    i