Background
Alibaba Cloud announced Qwen4-Max in a press release datelined Hangzhou, November 16, 2026.[1] The model is the flagship of the Qwen family, and Alibaba Cloud Model Studio lists the release date as 2026-11-16.[2]
What happened
Alibaba released Qwen4-Max as a sparse Mixture-of-Experts model with 2.4 trillion total parameters, activating 71 billion per token according to the company.[1] It has a one-million-token context window.[1] The model is available now through QwenCloud's Token Plan and API, and across QwenWork and QwenStudio.[1]
Key details
Alibaba's company-reported results include 6th place on Code Arena, a real-world coding leaderboard.[1] On Alibaba's Work360 evaluation, Qwen4-Max uses 35% fewer tokens than Qwen3.8-Max at High effort, and the company says average cost per task falls by about 60%.[1]
Alibaba also reported a 48-hour continuous agent run with Qwen4-Max that produced an experimentally validated coolant molecule.[1]
Open weights
Alibaba does not describe Qwen4-Max as open-weight. The company says smaller open-weight Qwen4 models will be released soon.[1]
Impact
The release gives developers and customers a new flagship option within the Qwen family, reachable through Alibaba's cloud services and applications. Its figures on token use and cost per task are the company's own, and they are the basis on which Alibaba presents the model's efficiency.
Sources
- 1
- 2Newly released models (Alibaba Cloud Model Studio)Alibaba Cloud
Cite this page: Alibaba releases Qwen4-Max, a 2.4-trillion-parameter MoE flagship. TellWorld, https://www.tellworld.net/news/qwen4-max-2026. Updated 2026-11-18