Alibaba Group has released the latest version of its Tongyi Wanxiang model series, unveiling the Wanxiang 2.6 lineup with significant upgrades aimed at professional film, television and image creation.

Alibaba said Wanxiang 2.6 delivers improved visual quality, audio performance and instruction-following capabilities, and supports video generation of up to 15 seconds per clip, the longest single-shot duration currently available from domestic models in China.

The new version also introduces role-playing and storyboard control features, enabling one-click generation of videos involving single or multiple characters, as well as interactions between people and objects. The model can automatically handle multi-shot transitions and other professional-grade tasks, catering to cinematic production needs.

Wanxiang 2.6 has been launched simultaneously on Alibaba Cloud’s Bailian platform and the Wanxiang official website. The broader Wanxiang model family now supports more than 10 visual creation functions, including text-to-image, image editing, text-to-video, image-to-video, voice-driven video generation, motion generation, role-playing and general video editing, and has been widely applied in AI comics, advertising design and short-form video production.