Alibaba launches video model Wan3.0 supporting document-to-video conversion
创业邦 · wechat · 2026-08-25
Alibaba officially launched its video generation large model, Wan3.0. The model supports generating 30-second videos in a single go and introduces the ability to input documents directly (including doc, xls, ppt, pdf, md formats) for video generation, aiming for high-fidelity real-world simulation. It is available on Alibaba Cloud Bailian and Qianwen platforms with limited-time API discounts. Additionally, Feixiang Teacher 3.0 was released, allowing one-click generation of runnable educational apps; Harvey, the highest-valued legal AI unicorn, launched its proprietary model Tenet based on Kimi K3; and Moonshot AI announced that Kimi K2.5 will be decommissioned at the end of this month.
Related event: Alibaba Launches Wan 3.0: 30-Second Native Video with Audio Generation(14 posts)→
More from Multimodal
- Fun video: AI blends cats, dogs, and fish into bizarre creatures — littleteckmonkey · 2026-08-25
- AI music experiment: DJ FL-AI releases new track "I am AI" — AIandDesign · 2026-08-25
- Magnific Launches Wan 3.0: Consistent Character Generation Across 22 Years — iamfakhrealam · 2026-08-25
- TLive-Omni: Omni-Modal Model for E-Commerce Live Streaming Understanding — TaoLiveAIGC · 2026-08-25
- Alibaba Wan 3.0 launches on Merge: 30s 1080p video in one pass — shensi · 2026-08-25
- Which model provides the utmost in fine detail for landscapes? — Dekes1 · 2026-08-25