Glonce

Alibaba's Wan3.0 doubles AI video length to 30 seconds

modelsThe Glonce editorsThe Decoder🔥70Glonce rating70Impact68Novelty62Relevance98Surprise35Players75Credibility82
Alibaba's Wan3.0 doubles AI video length to 30 seconds

Alibaba's video generation model Wan3.0 is now available in beta, producing AI videos up to 30 seconds long, double the maximum length of its predecessor, Wan2.5. The model processes text, images, video, and audio at the same time: a single prompt can include up to ten images, five videos, and five audio clips. Web pages and documents such as PDFs and PowerPoint presentations also work as input, turning static data into video. Wan3.0 recommends a video length based on the user's prompt and includes a tool for extending videos that already exist.

AI-generated video commonly suffers from visual drift and distortion, especially in faces and user interfaces. According to Alibaba, Wan3.0 is designed to fix that by keeping details from reference material, such as characters, props, and spatial layouts, more consistent across a generated video.

Wan3.0 is available through the wan.video website, Alibaba Cloud Model Studio, or via API on Qwen Cloud. It comes in two tiers: a Standard version, currently offered at a 30 percent discount, and a faster Prime version. No dollar pricing has been disclosed for either tier.

Alibaba is pitching Wan3.0 for a wide range of uses: speeding up film production, generating short dramas and social media clips for content creators, turning text and images into marketing and training videos for businesses, and producing realistic simulation footage to train autonomous vehicles and robotics systems.

The launch comes as Alibaba ramps up AI spending. The company just announced the largest share sale by a Hong Kong-listed company to fund its AI push, and the week before the Wan3.0 launch it reported a 75 percent year-over-year drop in quarterly profit, driven by sharply higher AI investment.

Key facts

  • Wan3.0 produces AI videos up to 30 seconds long, double the maximum length of predecessor Wan2.5.
  • A single prompt can combine up to ten images, five videos, and five audio clips, plus web pages and documents such as PDFs and PowerPoint files as input.
  • Alibaba says Wan3.0 keeps characters, props, and spatial layouts more consistent, addressing the visual drift and distortion common in AI-generated video.
  • Wan3.0 is available via wan.video, Alibaba Cloud Model Studio, and Qwen Cloud's API, in a Standard tier (currently 30 percent off) and a faster Prime tier.
  • The release follows Alibaba's largest share sale by a Hong Kong-listed company, aimed at funding AI spending, after a 75 percent year-over-year drop in quarterly profit.

Why it matters

Wan3.0 extends Alibaba's video generation line on two fronts at once. Length: the new 30 second maximum is double what Wan2.5 produced. Input flexibility: a single prompt can now combine text, up to ten images, five videos, and five audio clips, plus documents such as PDFs and PowerPoint files and even web pages. Turning static documents directly into video is a new capability for the line, not just a longer runtime.

Who it affects

Alibaba names several groups it built Wan3.0 for: film and short-drama producers and social media content creators looking to speed up production, businesses that want to turn text and images into marketing or training videos, and developers who need realistic simulation footage to train autonomous vehicles and robotics systems.

How to use it

Wan3.0 is available now, in beta, through the wan.video website, through Alibaba Cloud Model Studio, or via API on Qwen Cloud. It ships in two tiers: a Standard version, currently discounted 30 percent, and a faster Prime version; Alibaba has not disclosed dollar pricing for either one. The model also recommends a video length for a given prompt and includes a separate tool for extending videos that already exist.

How solid is it

Every claim in the report (the length increase, the input limits, the consistency improvements) is attributed to Alibaba itself or to the article, not to any named executive or researcher outside the company. The source gives no exact date for the beta launch and makes no comparison to competing video-generation models.

Risks and caveats

Alibaba's claim that Wan3.0 is designed to fix the visual drift and distortion common in AI-generated video, especially in faces and user interfaces, by keeping characters, props, and spatial layouts more consistent, is the company's own characterization, not an independently verified result. No dollar pricing is given for the Standard or Prime tier, only the 30 percent discount on Standard, and no dollar amount is disclosed for the Hong Kong share sale mentioned alongside the launch. That share sale and a 75 percent year-over-year drop in quarterly profit, both tied to Alibaba's rising AI spending, mean Wan3.0 is launching while the company's own AI investment is squeezing its bottom line.