Alibaba’s Qwen4-Max arrives as a preview on Qwen Chat

Alibaba on November 11 put Qwen4-Max-Preview on its Token Plan, where the model debuted first, with access also in Qwen Chat and Agent Mode, per a post on Threads. If your team wants to try Qwen4 on coding or agent work, those three surfaces are the way in. It is still a preview, though, and no benchmarks are out yet, so everything said about its quality comes from Alibaba itself.

Developers have already had a look under the hood. Alibaba called the open-weights Qwen3.8-Flash-Next “an early preview of the architecture used in Qwen4”, so the design reached local hardware before any model carrying the Qwen4 name did. Today’s preview is the first one under that name.

Alibaba’s pitch offers “an early look at Qwen4’s deeper knowledge, richer world understanding, and stronger coding and agentic capabilities.” The company lists four target workloads for the model: complex coding, office work, front-end design and professional software tasks.