Skip to content

MMAnswerBench

A multimodal mathematical reasoning benchmark that tests whether models can answer visually grounded math questions correctly.

10 models

Higher is better.

  1. 1GLM-5.2Z.AI91.0
  2. 2Kimi K2.6Moonshot AI86.0
  3. 3Claude Opus 4.5Anthropic84.0
  4. 4GLM-5.1Z.AI83.8
  5. 5Qwen3.6 PlusAlibaba83.8
  6. 6GLM-5Z.AI82.5
  7. 7Kimi K2.5Moonshot AI81.8
  8. 8Qwen3.5 397BAlibaba80.9
  9. 9Qwen3.6-27BAlibaba80.8
  10. 10Qwen3.6-35B-A3BAlibaba78.9
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Visual and structured mathematical QA
Tasks
Multimodal math questions
Authors
Qwen (2026)

Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.