Alibaba has rolled out a beta version of the latest Wan3.0 AI model that brings support for 30-second clip lengths and a broad range of multimodal reference inputs. The model combines high-fidelity video creation, precise visual consistency, and multimodal input handling into a single tool aimed at streamlining professional creative workflows.
The main upgrade centers around the extension of video length, where mainstream services that also includes the currently public Wan2.7-Video, are fixed at 16 seconds. With Wan3.0’s 30 seconds offering, there’s more flexibility for creators to execute more complex camera movements and longer, unbroken shots. The model also adds an intelligent duration feature that recommends an optimal video length based on the user’s prompt, along with video extension tools for expanding narrative timelines further.
The multimodal input range also allows the model to process process text, image, video, and audio simultaneously, and also accepts web pages and documents like PDFs and PowerPoint files, effectively letting users turn static, text-heavy material directly into dynamic video.
Wan3.0 focuses on high-precision visual continuity, rendering realistic human faces with synchronized micro-expressions, generating natural multilingual voice output, and keeping software interfaces and motion graphics stable throughout a clip. Beyond general visual stability, the model is built to replicate fine details from reference inputs closely, maintaining strict control over character and product details, spatial relationships, and voice consistency rather than producing rough approximations, while pairing that precision with natural movement and emotional expression.
Users can now apply to test the model through Alibaba Cloud’s Model Studio and its Qwen Cloud AI-native platform as public beta testing gets underway.
