Chinese tech giant Alibaba has introduced Wan 2.1, the latest version of its AI-powered video and image generation model, making it available as open-source. This move aims to expand access to advanced AI technology and foster innovation within the global developer community.
Wan 2.1 is capable of generating high-quality images and videos based on textual or graphical descriptions. The model supports both Chinese and English, making it a versatile tool for users worldwide. According to Alibaba, Wan 2.1 ranks among the top models on the VBench leaderboard, which evaluates video generation models based on factors like motion dynamics, spatial relationships, and object interactions.
Alibaba has released four versions of the Wan 2.1 model, featuring different parameter sizes—14 billion and 1.3 billion. These models are available for download on ModelScope and Hugging Face, allowing researchers, academic institutions, and commercial organizations to integrate them into their projects.
The launch of Wan 2.1 highlights Alibaba’s commitment to transparency and collaboration in the AI sector, providing the community with powerful tools for high-quality visual content creation and driving innovation forward.