Alibaba has made its video- and image-generating AI model, Wan 2.1, publicly available, marking a significant step towards increasing its adoption and intensifying competition in the AI sector.

This move follows a similar action by startup DeepSeek, which garnered attention for its low-cost open-source models earlier this year. These models have demonstrated performance comparable to well-established rivals like OpenAI, surprising many in the capital-intensive AI industry.

Key Details:

• Models Released: Alibaba has released four variants of Wan 2.1 - T2V-1.3B, T2V-14B, I2V-14B-720P, and I2V-14B-480P. These models can generate images and videos from text and image input, with the "14B" variants capable of processing 14 billion parameters for more accurate results.

• Availability: The models are available globally on Alibaba Cloud's ModelScope and HuggingFace platforms for academic, research, and commercial purposes.

Background:

In January, Alibaba introduced the latest version of its AI model, initially named Wanx, before shortening it to Wan. The model is known for generating highly realistic visuals and has secured top rankings on VBench, a leaderboard for video generative models, particularly excelling in multi-object interaction functionality.

Future Plans:

Alibaba also previewed its reasoning model QwQ-Max, which is expected to be made open source upon full release. Additionally, the company announced plans to invest at least 380 billion yuan ($52 billion) over the next three years to enhance its cloud computing and AI infrastructure.

Conclusion:

Alibaba's decision to make Wan 2.1 open source is likely to boost its utilization and drive further advancements in AI technology, positioning the company as a formidable player in the AI landscape.