Skip to content

SimpleVQA

A visual question answering benchmark focused on straightforward image-grounded understanding.

17 models

Higher is better.

  1. 1Qwen3.7 PlusAlibaba81.7
  2. 2Step 3.7 FlashStepFun79.2
  3. 3Qwen3.8 MaxAlibaba75.0
  4. 4Muse Spark OpenMeta73.6
  5. 5Gemini 3 ProGoogle73.2
  6. 6dots3-note PreviewDots Studio72.5
  7. 7Gemini 3.1 ProGoogle72.4
  8. 8Muse SparkMeta71.3
  9. 9Kimi K2.5Moonshot AI71.2
  10. 10Qwen3.5 397BAlibaba67.1
  11. 11Claude Opus 4.5Anthropic65.7
  12. 12GPT-5.4OpenAI61.1
  13. 13Qwen3.6-35B-A3BAlibaba58.9
  14. 14Grok 4.20xAI57.4
  15. 15Qwen3.6-27BAlibaba56.1
  16. 16GPT-5.2OpenAI55.8
  17. 17LFM2.5-VL-3BLiquid AI35.4
Higher is better. Bar colour: blue proprietary, orange open weights, green other.

About this benchmark

Metric
Image-grounded question answering
Tasks
Visual QA tasks
Authors
Z.AI (2026)

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