Wikibench

MM-MT-Bench

multimodal

A multi-turn LLM-as-a-judge evaluation benchmark for testing multimodal instruction-tuned models' ability to follow user instructions in multi-turn dialogues and answer open-ended questions in a zero-shot manner.

Leaderboard

Showing 17 of 17 results

  1. Mistral Large 3
    84.9
    i
  2. Pixtral Large
    74
    i
  3. Pixtral-12B
    60.5
    i
  4. Qwen3 VL 235B A22B Instruct
    8.5
    i
  5. Qwen3 VL 235B A22B Thinking
    8.5
    i
  6. Qwen3 VL 32B Instruct
    8.4
    i
  7. Qwen3 VL 32B Thinking
    8.3
    i
  8. Qwen3 VL 30B A3B Instruct
    8.1
    i
  9. Qwen3 VL 8B Thinking
    8
    i
  10. Qwen3 VL 30B A3B Thinking
    7.9
    i
  11. Qwen3 VL 4B Thinking
    7.7
    i
  12. Qwen3 VL 8B Instruct
    7.7
    i
  13. Qwen3 VL 4B Instruct
    7.5
    i
  14. MiniStral 3 (14B Instruct 2512)
    0.085
    i
  15. Ministral 3 (8B Instruct 2512)
    0.081
    i
  16. Ministral 3 (3B Instruct 2512)
    0.078
    i
  17. Qwen2.5-Omni-7B
    0.06
    i