Alibaba launched the video model Wan3.0 with support for documents and presentations

Chinese Alibaba has released Wan3.0 — a new version of its AI model for video generation. The release came just two days after the company raised $10.2 billion through a secondary stock offering on the Hong Kong Stock Exchange, marking the largest secondary offering in the exchange's history.

Public beta testing of Wan3.0 began on August 6 on Alibaba Cloud Model Studio and Qwen Cloud platforms, and the model is now available to a broader audience. Wan3.0 generates videos with a resolution up to 1080p and a duration of up to 30 seconds — twice as long as the previous version, Wan2.7, which was limited to 15 seconds. The model maintains character details, object appearance, spatial composition, and animation graphics throughout the entire video, synchronizes facial expressions, and adds voiceovers in multiple languages.

In addition to text, images, audio, and video, Wan3.0 accepts web pages and documents as input — including files in PDF, PowerPoint, and Excel formats. Users can upload a presentation or spreadsheet and receive a video that reflects the structure of the original material. According to Alibaba, the model is already being used in the production of short dramas and films, advertising and marketing, tourism promotion, music video creation, as well as in simulation training for autonomous vehicles and robotics.

The cost of video generation in Wan3.0 is $0.05 per second at 480p resolution, $0.1 per second at 720p, and $0.2 per second at 1080p — one minute of video in maximum quality will cost $12. For comparison, Google Veo 3.1 charges $0.4 per second of video. From August 24 to September 23, a temporary 30% discount on Wan3.0 API prices is available on selected platforms.

Source: x.com