Alibaba on Monday launched Wan3.0, a new AI video generation model capable of producing native 30-second clips at up to 1080p resolution, just one day after the Chinese tech giant closed a $10.2 billion share offering to fund its artificial intelligence infrastructure expansion.
The model, available through Alibaba Cloud Model Studio, represents a significant step forward in AI-generated video. Wan3.0 can generate complete audio-visual content from text, images, audio, video, documents, and even web pages as input, combining up to 10 images, 5 videos, and 5 audio clips in a single generation pass, according to Reuters.
30-Second Native Generation
Unlike earlier models that stitched together shorter segments, Wan3.0 generates up to 30 seconds of continuous video in a single pass. The model supports native audio-visual generation, meaning dialogue, sound effects, and music are produced alongside the visual content rather than added in post-production. Alibaba positions this as a tool for advertising, film production, and content creation at scale.
Pricing is structured by duration and resolution. At the current 30% introductory discount valid until September 24, 480p output costs $0.035 per second, 720p costs $0.07, and 1080p costs $0.14, making a full 30-second clip at maximum resolution roughly $4.20. A faster Prime tier is also available for time-sensitive production work, as detailed on Quartz.
Funding the AI Push
The launch comes amid Alibaba aggressive spending on AI. The company closed a $10.2 billion share placement on August 23, one of the largest equity raises in Chinese tech history, to finance data center expansion, model training, and talent acquisition. The capital raise signals Alibaba determination to compete with Microsoft, Google, and Amazon in the global AI race despite mounting competition from Chinese rivals including ByteDance and Baidu.
The model is now generally available to all developers through the Alibaba Cloud API, with no application required. Alibaba has also released an experience center where users can test the model before integrating it into production workflows. The company claims the model achieves strong character consistency, realistic facial rendering, and precise lip-syncing across multiple languages.
Alibaba small-model division also made news this week with a model capable of running on a laptop while matching cloud-scale performance, suggesting the company is pursuing both ends of the compute spectrum simultaneously. For the broader AI industry, Wan3.0 competitive pricing and multimodal capabilities intensify pressure on Western rivals to match Chinese cost structures.
discussion