Alibaba Releases Over 100 New Open-Source AI Models and an AI Video Tool

The Chinese tech giant Alibaba released over 100 new open-source AI models and a text-to-video AI tool this Thursday. According to Reuters, the tech company is increasing its efforts with new AI technologies to compete in the AI market.

The new AI models are from the Qwen suite, which is the main open-source AI model developed by Alibaba’s research division, launched in May this year. The latest model versions, the Qwen 2.5, range from 0.5 to 72 billion parameters and can be used in more than 29 languages.

These new models can support multiple AI applications in different sectors including gaming, automobile, and scientific research, and can be accessed from anywhere and by everyone across the globe. According to CNBC, these LLMs have advanced math and coding capabilities and can generate text and images from prompts.

The Chinese company also said that the newest version of its flagship model Qwen-Max—which is not open source—is superior to Meta and Open AI’s models in language comprehension, reasoning, and other areas.

As part of its Tongyi Wanxiang image products, Alibaba also launched its new text-to-video model, an AI tool that creates videos from prompts just like OpenAI’s Sora. More companies have recently joined this AI video sector. Just a few days ago Adobe announced the development of a similar tool, the Firefly Video Model, and Tiktok’s parent company ByteDance launched its AI video tool Jimeng AI last month.

“Alibaba Cloud is investing, with unprecedented intensity, in the research and development of AI technology and the building of its global infrastructure,” said Alibaba’s CEO Eddie Wu.