Alibaba Will Launch A Deep Reasoning Model Based on Qwen2.5-MAX
Alibaba Group Holding Limited CEO Eddie Wu stated at the financial report meeting for the third quarter of the 2025 fiscal year that Alibaba will soon release a deep reasoning model based on Qwen2.5-MAX.
On January 29th, during the New Year celebration, Alibaba Cloud announced its new large-scale MoE model called Qwen 2.5-Max, claiming to have surpassed competitors like DeepSeek V3 in multiple benchmark tests.
Alibaba Cloud later announced that Qwen2.5-Max surpassed models such as DeepSeek-V3, Open AI o1-mini, and Claude-3.5-Sonnet in the Chatbot Arena large-scale blind test, ranking seventh globally with a score of 1332. It is also the champion among Chinese large models in non-inferential categories. At the same time, Qwen2.5-Max ranks first in individual abilities such as mathematics and programming, and second in hard prompts.
SEE ALSO: Alibaba’s Cloud and AI Investment in the Next Three Years Expected to Outpace Last Decade
