Alibaba announced Wednesday that it is making its advanced artificial intelligence models for video generation available for free use, intensifying competition with industry leaders such as OpenAI.
The Chinese tech giant is open-sourcing four models from its Wan2.1 series, the newest iteration of their foundational AI, which can produce images and videos from text and image inputs. These models will be accessible through Alibaba Cloud’s Model Scope and Hugging Face, a widely used repository for AI models, available to academics, researchers, and businesses worldwide.
Open-source AI has gained significant attention recently. This approach contrasts with proprietary models like those developed by OpenAI. In January, DeepSeek, another Chinese firm, shook up global markets with its open-source AI model, claiming to have trained it at a lower cost using less advanced Nvidia chips compared to other major AI players. Like Alibaba’s, DeepSeek’s model is open source, enabling others to download and modify its code.
Open sourcing tech serves multiple strategic goals, including fostering innovation and creating a community around the product. A lively discussion is currently underway about the potential for AI models to become widely available and commoditized.
Chinese companies are particularly active in promoting open-source models, with Alibaba and DeepSeek’s models gaining global popularity. Alibaba launched its first open-source model in August 2023, while Meta is leading the open-source initiative in the U.S. with its Llama models.
Alibaba’s stock has performed strongly this year, with its Hong Kong listing up 66% in 2025 to date. This success has been fueled by various factors, including the company’s improved financial results, its standing as a significant AI player in China, and recent indications of support from Chinese President Xi Jinping for the domestic private sector.
