Vision2Web
A benchmark for converting visual references into functional web implementations.
4 models
Higher is better.
- 1Qwen3.8 MaxAlibaba69.0
- 2Qwen3.8-Flash-NextAlibaba64.0
- 3Qwen3.8-27BAlibaba62.9
- 4Qwen3.8-Omni-FlashAlibaba62.9
About this benchmark
- Metric
- Visual reference to web implementation
- Tasks
- Screenshot-to-web tasks
- Authors
- Z.AI (2026)
- Source
- GLM-5V-Turbo ↗
Scores are as reported by model developers or evaluators and may use different settings (for example reasoning effort or tools). Compare with care.