To further support the open-source community, Alibaba Cloud has freely opened up a total of 4 selections to the public to download, explore, and further cultivate its video-generating AI models.
The models in question are from the Wan2.1 series, including 14B and 1.3B of Tongyi Wanxiang (Wan,) which has surpassed 1M combined downloads on the ModelScope platform and Hugging Face Hub. Designed to generate high-quality images and videos from both text and image inputs, the actual models are T2V-14B, T2V-1.3B, I2V-14B-720P, and I2V-14B-480P.
Unveiled earlier this year, the Wan2.1 series is the first video generation model to support text effects in both Chinese and English while capable of producing realistic visuals by accurately handling complex movements, improving pixel quality, adhering to physical principles, and optimizing instruction execution precision.
After securing the top spot on the VBench leaderboard, which is a benchmark suite for video generative models, Alibaba Cloud hopes that these open-source models could lower entry barriers for businesses looking to leverage AI for high-quality visual content creation in a cost-effective manner. By providing a 14B and 1.3 option separately, users can find out the one that matches both hardware and output requirements.
