At this year’s Apsara Conference, the Alibaba Group presented its largest open-source initiative to date, with a focus on the development of artificial intelligence (AI). The company released more than 100 new large language models (LLMs) that are available as open source. These models, known as Qwen 2.5, are expected to be widely used in various sectors – from the automotive industry to video games and scientific research.

Alibaba Cloud, the technological backbone of the group, also presented a new text-to-video model. This technology, part of the Tongyi Wanxiang LLM family, enables the generation of high-quality videos with a resolution of 1080p. These videos can be created in a variety of visual styles, including 3D animations and realistic scenes, and support both Chinese and English text prompts. This puts Alibaba in the global race for the leading role in the emerging text-to-video technology.

The increasing pace of AI development demonstrates the company’s ambition to create an advanced AI infrastructure. Alibaba expects that most physical things in the future will be equipped with AI capabilities. The transition to artificial general intelligence (AGI) – software with human-like intelligence and learning abilities – is seen as a long-term goal.

As part of its open-source strategy, Alibaba has drastically reduced the prices for using its LLMs, with the company announcing a price cut of up to 85 percent for key AI models. This reflects the company’s commitment to making AI developments widely accessible and promoting innovation.

Image> Alibaba