BEIJING, China: Alibaba has made its video- and image-generating AI model, Wan 2.1, publicly available, marking a significant step in the growing competition within artificial intelligence.

The move aligns with a broader industry trend of open-sourcing AI models to increase accessibility and adoption.

The Chinese e-commerce giant headquartered in Hangzhou, announced this week that four versions of Wan 2.1—T2V-1.3B, T2V-14B, I2V-14B-720P, and I2V-14B-480P—are now available on Alibaba Cloud's ModelScope and HuggingFace. These models allow users to generate high-quality images and videos from text and image prompts, with 14B models processing up to 14 billion parameters for enhanced accuracy.

This decision comes amid a surge in open-source AI models, with startups like DeepSeek gaining traction for offering cost-effective alternatives to major players like OpenAI. By making Wan 2.1 available to academic, research, and commercial users, Alibaba aims to increase adoption while reinforcing its position in AI-driven content creation.

Alibaba first introduced Wanx 2.1 in January, later rebranding it as Wan, emphasizing its ability to produce highly realistic visuals. The company has since showcased top rankings on VBench, a leaderboard evaluating video generative models, particularly for multi-object interactions—a key capability in improving AI-generated motion realism.

Beyond image and video generation, Alibaba is also expanding in AI reasoning. A day before the launch, it previewed QwQ-Max, an advanced reasoning model, with plans to make it open source upon full release.

Alibaba is backing its AI ambitions with major investment plans. The company has pledged to invest at least 380 billion yuan (US$52 billion) over the next three years to strengthen cloud computing and AI infrastructure.