Alibaba Challenges DeepSeek: The New AI Qwen 2.5 Shifts the Balance of Power

The Chinese tech company Alibaba (9988.HK) made a surprising announcement by unveiling the latest version of its artificial intelligence model, Qwen 2.5-Max, on the first day of the Lunar New Year. This move signals increasing pressure from the Chinese startup DeepSeek, which has made significant breakthroughs in the AI field over the past weeks, as reported by NewsBox.

According to an official statement from Alibaba's cloud division, Qwen 2.5-Max surpasses leading models, including GPT-4o, DeepSeek-V3, and Llama-3.1-405B, in most key metrics. The release of Qwen 2.5 confirms Alibaba's serious intent to strengthen its position in the rapidly evolving AI market, driven by the influence of new players, writes Reuters.

On January 10, the Chinese startup DeepSeek launched its model, DeepSeek-V3, and on January 20, it introduced an even more powerful version, R1. This caused quite a stir in Silicon Valley, leading to a decline in the stocks of tech giants as DeepSeek's low development and operational costs compelled investors to reevaluate the strategies of major American companies in the AI sector.

Nevertheless, DeepSeek's rapid success spurred Chinese competitors to accelerate the modernization of their own models. ByteDance, the owner of TikTok, presented an updated version of its flagship model just two days after DeepSeek-R1 was released. According to ByteDance, this new version outperformed OpenAI o1 in the AIME benchmark test. This development was a direct response to DeepSeek's claims about the competitiveness of its R1 model compared to OpenAI o1.