Wan 3.0 is characterized by its unique ability to directly transform raw data files and office documents into consistent video clips with extreme precision and competitive economic efficiency.
Article Index:
- Launch of Wan 3.0 and record-breaking Hong Kong share offering
- Technical capabilities and video generation from office documents
- Pricing structure and closed model strategy
- Financial performance and exceptional cloud revenue growth
- Frequently asked questions
Launch of Wan 3.0 and record-breaking Hong Kong share offering
Chinese tech giant Alibaba announced on Monday the official launch of the latest version of its AI video generation model, “Wan 3.0.” This advanced system is capable of producing complete cinematic video clips lasting up to 30 seconds directly from text documents, spreadsheets, presentation slides, and web pages. This commercial rollout comes just one day after the company announced the completion of a secondary share offering on the Hong Kong Stock Exchange valued at approximately $10.2 billion USD. This marks the largest secondary public offering ever recorded by a company listed in Hong Kong, with all financial proceeds allocated to fund AI infrastructure investments.
Technical capabilities and video generation from office documents
The “Wan 3.0” model doubles the duration of the clips generated by its predecessor, “Wan 2.7,” producing 30-second footage in a single processing pass at up to 1080p resolution. Alibaba emphasizes that the model maintains meticulous character details, spatial layouts of elements, and motion consistency across the entire scene, while generating precisely synchronized facial expressions and multilingual audio tracks.
The model’s most unique feature lies in the diversity of inputs it accepts. In addition to text, images, audio recordings, and videos, “Wan 3.0” can read and process document files in various formats, spreadsheets, presentation slides, PDFs, and formatted text. This allows users to provide a complete presentation or spreadsheet and receive an organized, explanatory video in return. This system serves as an ideal tool for corporate communications, marketing, short-drama production, and interactive content, as well as generating visual simulation environments for training autonomous driving systems.
Pricing structure and closed model strategy
Alibaba has set competitive pricing for the model starting at $0.05 per second for 480p resolution, $0.10 for 720p, and $0.20 for 1080p—equivalent to $12 per minute for the highest tier. This is significantly lower than Google’s “Veo 3.1” model pricing of $0.40 per second. A 30% promotional discount on APIs is available until September 23. The platform’s preview version has been available since August 6 via Alibaba Cloud Model Studio and the “Qwen” platform, with plans to launch a dedicated subscriber platform to manage computing costs. Notably, the model weights remain closed, marking a shift from the open-source approach the company followed with its “Wan 2.2” release in July 2025.
Financial performance and exceptional cloud revenue growth
This move aligns with the company’s financial results, as cloud and AI revenues jumped 45% year-over-year to reach 48.44 billion yuan in the quarter ending in June, while model service revenues surpassed the 16 billion yuan annual recurring revenue mark. Conversely, net profit dropped by 75% as capital expenditures rose 75% to 67.68 billion yuan to support the company’s plan to inject up to 480 billion yuan into smart infrastructure.
Frequently Asked Questions
Question: What are the core features of the new Wan 3.0 model?
Answer: It generates 1080p video clips lasting 30 seconds directly from presentation files, spreadsheets, and text documents with superior motion consistency.
Question: How do Alibaba’s model prices compare to competitors like Google?
Answer: The top tier costs $0.20 per second ($12 per minute), which is half the cost of Google’s Veo 3.1 model at $0.40 per second.
Question: What is the scale of funding Alibaba raised to support AI projects?
Answer: It raised approximately $10.2 billion through additional share offerings in Hong Kong, allocating the entire amount to support computing infrastructure and artificial intelligence.
