Alibaba Cloud launches Wan3.0 AI video model with document support

Alibaba Cloud has officially released Wan3.0, its latest generative AI video model capable of producing clips up to 30 seconds in length. This update doubles the maximum duration of its predecessor, Wan2.7, which was limited to 15 seconds.

A key feature of Wan3.0 is its ability to generate video from diverse inputs, including text, images, audio, and existing video clips. Notably, the model supports document-based inputs, allowing users to convert files such as PDFs, PowerPoints, Excel spreadsheets, Word documents, and even live web pages into dynamic video content. The system can process files up to 100MB or 50 pages in length.

Technically, the model offers native audio generation, including synchronized voices and ambient sound, and supports resolutions up to 1080p. It is designed to maintain visual consistency for characters and environments across sequences.

The launch follows Alibaba’s recent $10 billion share offering aimed at funding its growing AI infrastructure and development. Alibaba Cloud has introduced a tiered API pricing structure based on resolution, with 1080p costs approximately $0.20 per second. To encourage adoption, the company is offering a 30% discount on API usage through September 23, 2026.