Alibaba’s Qwen2.5-Max is a large-scale Mixture-of-Expert (MoE) model that can achieve “competitive performance against the top-tier models,” WSJ reported citing a company statement.

Qwen2.5-Max is ranked seventh overall in ... This advanced LLM from Alibaba Cloud is a large-scale Mixture of Experts (MoE) model that was pretrained on over 20 trillion tokens and further ...

The S1 reasoning model was developed on top of the Chinese e-commerce giant’s Qwen2.5-32b-Instruct model by researchers from Stanford University, where Li works, and the University of Washington ...