Alibaba Cloud has launched its Qwen 2.5 multimodal models, featuring up to 72 billion parameters and supporting 29 languages, specifically designed for industries such as automotive and gaming.

In addition to the new models, Alibaba introduced a text-to-video tool that generates videos from text prompts, similar to OpenAI's Sora.

These advancements showcase enhanced knowledge, stronger programming capabilities, and improved mathematical skills within the new model series.

Eddie Wu, CEO of Alibaba Cloud Intelligence, highlighted the company's significant investment in AI technology and infrastructure development, including the unveiling of a next-generation data center architecture.

The new architecture, CUBE DC 5.0, is designed to improve energy efficiency and reduce deployment times by up to 50%.

This modular approach is expected to streamline data center deployment, allowing for quicker setups compared to traditional methods.

Alibaba also launched its 9th Generation Enterprise Elastic Compute Service (ECS), which boasts a 30% increase in search recommendation speed and enhanced query performance.

The introduction of open-sourced models is anticipated to save developers significant time and costs, while promoting broader adoption of Alibaba's technology.

As the global race for AI dominance intensifies, Alibaba seeks to assert its position in the rapidly evolving AI landscape, particularly in competition with US firms.

To further enhance its market position, Alibaba has reduced prices on its Qwen-Turbo LLM by 85%, along with price cuts on two other models.

Following these announcements, Alibaba's stock has risen over 4%, recovering from a challenging start to the year and increasing by more than 22% since late June.

These developments were showcased at the Apsara Conference, which runs from September 19 to 21 and features around 400 forums focused on AI and cloud computing.