Qwen 4 Max API quietly launches with major CoT reduction and intelligence density gains

Xianbao_QIAN · x · 2026-11-16

Qwen 4 Max's API is now live. The poster observes a big "intelligence density" win — noticeably shorter chain-of-thought tokens for the same capability, likely from new training data or methods. The smaller variants coming soon are the real story: if trained the same way, a new 27B model could be turbo-charged via domain adaptation and local deployment.

Related event: Alibaba Releases Qwen4-Max with Native Vision and 1M Token Context(21 posts)→

Original post →

More from Models

Models channel →