Wikibench

MMStar

reasoning official site →

Multilingual: no

MMStar is an elite vision-indispensable multimodal benchmark comprising 1,500 challenge samples meticulously selected by humans to evaluate 6 core capabilities and 18 detailed axes. The benchmark addresses issues of visual content unnecessity and unintentional data leakage in existing multimodal evaluations.

Methodology

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

Leaderboard

Showing 20 of 22 results

  1. Qwen3.6 Plus self-reported llm-stats
    83.3%
  2. Qwen3.5-122B-A10B self-reported llm-stats
    82.9%
  3. Qwen3.5-35B-A3B self-reported llm-stats
    81.9%
  4. Qwen3.6-27B self-reported llm-stats
    81.4%
  5. Qwen3.5-27B self-reported llm-stats
    81.0%
  6. Qwen3 VL 32B Thinking self-reported llm-stats
    79.4%
  7. Qwen3 VL 235B A22B Thinking self-reported llm-stats
    78.7%
  8. Qwen3 VL 235B A22B Instruct self-reported llm-stats
    78.4%
  9. Qwen3 VL 32B Instruct self-reported llm-stats
    77.7%
  10. Qwen3 VL 30B A3B Thinking self-reported llm-stats
    75.5%
  11. Qwen3 VL 8B Thinking self-reported llm-stats
    75.3%
  12. Qwen3 VL 4B Thinking self-reported llm-stats
    73.2%
  13. Qwen3 VL 30B A3B Instruct self-reported llm-stats
    72.1%
  14. Qwen3 VL 8B Instruct self-reported llm-stats
    70.9%
  15. Qwen2.5 VL 72B Instruct self-reported llm-stats
    70.8%
  16. Qwen3 VL 4B Instruct self-reported llm-stats
    69.8%
  17. Qwen2.5 VL 32B Instruct self-reported llm-stats
    69.5%
  18. Qwen2.5-Omni-7B self-reported llm-stats
    64.0%
  19. Qwen2.5 VL 7B Instruct self-reported llm-stats
    63.9%
  20. DeepSeek VL2 self-reported llm-stats
    61.3%