On August 31, Alibaba Qwen launched the Agent Teams feature, expanding the AI video creation method from primarily writing prompts to a multi-agent collaboration model where users propose ideas and Agents actively think, break down tasks, and execute them. The feature is now available on Qwen's PC and Qwen APP, supporting flexible collaboration across multiple devices.
At the model level, Qwen Creation has launched the Wan3.0 video generation model, and also introduced the enhanced Wan3.0Prime version, which generates faster and delivers better results. It also integrates top models such as Qwen-Image-3.0. These models are planned and called by Agent Teams according to different creative stages to produce outputs.
In terms of workflow, different role-based Agents such as screenwriters, directors, artists, and video generators collaborate to complete the entire process from idea planning, script writing, character setting, storyboard design, image generation, voiceover, music, to final production.
The application scenarios cover AI short dramas, comic dramas, marketing advertisements, image design, and game content: users propose story ideas, and Agents sequentially complete plot expansion, character setting, and final video production; after uploading brand materials and product samples, the system plans advertisement videos around target audiences and product highlights; in game scenarios, Agents can design character modeling, visual styles, and camera movements based on needs and support real-time adjustments to storyboards and final videos as ideas evolve.
