Alibaba Cloud has officially launched Wan3.0, its latest generative AI video model, which can produce videos up to 30 seconds long directly from inputs including documents, spreadsheets, presentations, PDFs, and web content.
Alibaba says the model delivers improved instruction-following, shot consistency, and audio quality compared with earlier versions. Wan3.0 entered public beta earlier this month and has already been tested for advertising, tourism, short-form entertainment, and music-video production use cases.
The launch arrives as Alibaba dramatically ramps up its financial commitment to AI: the company announced a roughly $10 billion share placement to help fund computing infrastructure and continued development of its Qwen model family and broader cloud business. That combination matters because generative AI competition is increasingly becoming a contest between companies that can simultaneously finance enormous computing requirements and keep shipping new, capable models at pace. Wan3.0 also illustrates a broader industry shift toward multimodal systems that treat existing business documents themselves as creative inputs — rather than requiring users to write carefully engineered prompts, platforms are beginning to let presentations, spreadsheets, and reports serve as the raw material for finished media, potentially widening AI adoption inside large corporations far beyond marketing and creative teams.
