WebMMU is a comprehensive benchmark designed to push the boundaries of AI for the web. It challenges models to answer questions about websites, edit real HTML/CSS/JS code, and generate web layouts from mockups—across four languages and 20+ domains. Whether you're building smarter web agents or testing the limits of multimodal models, WebMMU is your go-to testbed.
🌐 Multilingual: English, Spanish, German… See the full description on the dataset page:
https://huggingface.co/datasets/McGill-NLP/WebMMU.
