Alibaba launched Wan3.0 on Monday, expanding access to its latest video-generation model as the company ramps up spending on artificial intelligence.
The model can generate videos of up to 30 seconds in a single pass from text, images, audio, existing videos and documents including PDFs, presentations, spreadsheets and web pages, according to Alibaba Cloud. Reuters reported that the model had already been used in advertising, short dramas, tourism promotion and music video production since its public beta began Aug. 6.
For marketing, communications and training teams, the document support could reduce the work required to turn existing business material into video content.
Wan3.0 doubles the 15-second limit of its predecessor, Wan2.7, and supports resolutions of up to 1080p. Alibaba says Wan3.0 can preserve characters, objects, backgrounds and spatial layouts throughout a clip while producing synchronized facial expressions and multilingual voice output. These capabilities have not yet been independently verified.
The ability to process business documents could be one of Wan3.0's most commercially useful features.
Instead of creating a video script and visual assets from scratch, a marketing or communications team could provide a presentation, PDF or spreadsheet and use the model to turn that information into video content. That puts Wan3.0 firmly in the enterprise content-creation market alongside its more creative uses.
The model's pricing also gives Alibaba another way to compete. Wan3.0 is priced at 5 cents per second for 480p video, 10 cents for 720p and 20 cents for 1080p. Google’s Veo 3.1 standard tier, by comparison, costs 40 cents per second.
At the same time, price alone does not establish which model produces better results, and The Next Web noted that Wan3.0 has not yet been independently benchmarked.
Alibaba’s bigger AI bet
The launch follows Alibaba’s reported $10 billion share sale and comes a week after the company reported a 75% decline in quarterly profit. Capital expenditure rose 75% to 67.68 billion yuan (approximately $10.07 billion) during the period as Alibaba continued investing heavily in AI infrastructure. Alibaba reported that its cloud and AI revenue grew 45% to 44.84 billion yuan in the June quarter.
The sequencing — raise billions, then immediately ship a new product — is Alibaba's way of showing investors its AI spending produces something tangible, rather than asking them to take it on faith. The company has also committed 380 billion yuan (approximately $56.5 billion) over three years to AI infrastructure.
What comes next
Wan3.0 looks less like a pure creative experiment and more like a business productivity tool. Its ability to process documents and presentations could give Alibaba an opening with marketing, communications and training teams that need quick video without starting from scratch.
The tradeoff is access. Wider API availability is still rolling out, while the consumer-facing wan.video service is planned as a members-only platform. Alibaba has also kept Wan3.0’s weights closed, breaking from its earlier approach of open-sourcing Wan models.
That makes Wan3.0 an important test of whether Alibaba can turn rapidly improving AI capabilities into a profitable commercial business, especially after committing hundreds of billions of yuan to AI infrastructure.
Read more: Google’s Veo 3.1 turns photos into videos, offering another look at how major AI providers are competing on video-generation capabilities.