Wikibench

MathVista-Mini

math

MathVista-Mini is a smaller version of the MathVista benchmark that evaluates mathematical reasoning in visual contexts. It consists of examples derived from multimodal datasets involving mathematics, combining challenges from diverse mathematical and visual tasks to assess foundation models' ability to solve problems requiring both visual understanding and mathematical reasoning.

Leaderboard

Showing 20 of 23 results

  1. Kimi K2.5
    90.1%
    i
  2. Qwen3.5-27B
    87.8%
    i
  3. Qwen3.5-122B-A10B
    87.4%
    i
  4. Qwen3.6-27B
    87.4%
    i
  5. Qwen3.6-35B-A3B
    86.4%
    i
  6. Qwen3.5-35B-A3B
    86.2%
    i
  7. Qwen3 VL 32B Thinking
    85.9%
    i
  8. Qwen3 VL 235B A22B Thinking
    85.8%
    i
  9. Qwen3 VL 235B A22B Instruct
    84.9%
    i
  10. Qwen3 VL 32B Instruct
    83.8%
    i
  11. Qwen3 VL 30B A3B Thinking
    81.9%
    i
  12. Qwen3 VL 8B Thinking
    81.4%
    i
  13. Qwen3 VL 30B A3B Instruct
    80.1%
    i
  14. Qwen3 VL 4B Thinking
    79.5%
    i
  15. Qwen3 VL 8B Instruct
    77.2%
    i
  16. Qwen2.5 VL 72B Instruct
    74.8%
    i
  17. Qwen2.5 VL 32B Instruct
    74.7%
    i
  18. Qwen3 VL 4B Instruct
    73.7%
    i
  19. Qwen2-VL-72B-Instruct
    70.5%
    i
  20. Qwen2.5 VL 7B Instruct
    68.2%
    i