Alibaba's Wan3.0 turns text, images and files into 30s videos

Alibaba has officially launched Wan3.0: the video model generates clips up to 30 seconds long — double its predecessor Wan2.5. The standout is its inputs: besides text, it accepts up to 10 images, 5 videos and 5 audio clips, plus documents such as PDFs, web pages and PowerPoint decks, as TechNode and The Decoder report.

Alibaba Cloud highlights improvements in "instruction following, consistency across shots and audio quality." Consistency of characters, props and spatial layout is precisely the Achilles heel of generative video; Wan3.0 is designed to reduce visible drift and distortion.

Pricing is aggressive: a 30-second clip costs from $1.50 in 480p, $3 in 720p and $6 in 1080p (Standard mode); the Prime tier runs roughly 40 percent higher. A 30 percent discount applies on selected platforms from 24 August to 23 September. The model is available via wan.video, Alibaba Cloud Model Studio and the Qwen Cloud API.

The context reveals the ambition: Alibaba recently reported a 75 percent year-over-year drop in quarterly profit — the result of sharply higher AI spending, financed in part through a multi-billion dollar share sale. Wan3.0 is the first major product of that offensive.

For content teams, the implication is direct: a slide deck or product document becomes a finished clip via prompt — at prices far below traditional video production. The price war in AI video, where OpenAI, Google and Runway also compete, enters its next round.