Chinese e-commerce giant Alibaba (9988.HK) has announced that its latest video- and image-generating AI model, Wan 2.1, is now open-source. This move is expected to boost adoption and intensify competition in the AI sector.
Following a similar approach to startup DeepSeek, whose cost-effective open-source models made waves earlier this year, Alibaba’s decision marks a significant step in making advanced AI tools more accessible.
Alibaba has released four versions of Wan 2.1—T2V-1.3B, T2V-14B, I2V-14B-720P, and I2V-14B-480P. These models, which generate images and videos from text and image inputs, vary in capability, with the “14B” variants processing 14 billion parameters for enhanced accuracy.
Available on Alibaba Cloud’s ModelScope and Hugging Face, Wan 2.1 caters to academic, research, and commercial use. Originally introduced in January as Wanx before being rebranded, the model has gained recognition for generating highly realistic visuals and leading video generative model rankings on VBench.
Additionally, Alibaba has previewed QwQ-Max, an AI reasoning model set to be open-sourced upon full release. The company also unveiled plans to invest at least 380 billion yuan ($52 billion) over the next three years to strengthen its cloud computing and AI infrastructure.
