article --summary
The summary
Alibaba has formally launched Qwen4-Max, the first model in its Qwen4 family, pitched as an always-on AI collaborator for coding, research and other professional work. The model uses a mixture-of-experts design with a very large parameter count, accepts images as well as text, and offers a one-million-token context window.
The article covers Alibaba's own benchmark claims, which show better scores and lower token use and cost compared with the previous Qwen3.8-Max. It also notes that independent testing by Artificial Analysis places the model well behind several leading rivals, though it sits slightly above the median for its price class.
The piece then looks at Alibaba's open-weight history. Qwen3.8-Max's weights were released late, under a restrictive licence, and were a limited version of the hosted product. Qwen4-Max has no announced weights, leaving developers dependent on Alibaba's services, though smaller Qwen4 models are promised open release.
Editorial note: This is a summary of an article first published by The New Stack. The full piece is on their site.