Alibaba Unveils AI Model, Outperforming Rivals in Key Areas
Alibaba Group Holding announced the release of a new open-source artificial intelligence (AI) reasoning model on Thursday, claiming its performance exceeds that of DeepSeek’s R1. The Chinese technology giant’s announcement highlighted its advancements in AI capabilities, encompassing both models and data-center infrastructure.
Alibaba shares have surged in Hong Kong after it introduced its latest AI reasoning model. Photo: AFP
Following the launch of the QwQ-32B model, Alibaba’s Hong Kong-listed shares saw a significant increase, closing nearly 8.4 percent higher at HK$140.80 on Thursday. The South China Morning Post is owned by Alibaba.
Despite its relatively smaller size, with only 32 billion parameters, Alibaba’s new model achieved results comparable to or better than DeepSeek’s R1, which boasts 671 billion parameters. According to a blog post by the team behind Alibaba’s Qwen family of AI models, the QwQ-32B demonstrated superior performance in areas such as mathematics, coding, and general problem-solving.
The team emphasized that the QwQ-32B’s smaller parameter count allows it to operate with reduced computing resource requirements. This lean design enables wider adoption and aligns with Alibaba chairman Joe Tsai’s views, as expressed in a recent column for the Post. Tsai emphasized the importance of practical applications when developing AI models to maximize intelligence.
