China's Claude killer lands next week: Alibaba readies Qwen 4

Alibaba has put a date on Qwen 4. At the Apsara Conference in Amsterdam on October 29, Alibaba Cloud CTO Feifei Li said the next generation of models will ship as soon as next week, according to people in the room, including researchers from the Qwen team itself.

The whole lineup was on stage at once, four models. Qwen4-Max is the flagship built for top quality, Qwen4-Plus and Qwen4-Flash are meant to balance speed and price, and Qwen4-27B is sized to run on your own hardware. That last one matters most to regular users: Qwen has long been one of the most popular open models people install on a home computer, with no subscription and no dependence on someone else's servers.

Qwen 4 itself is not a surprise. In September, at Alibaba's main conference in Hangzhou, the company said the model was in training and laid out what comes next: Qwen 4.5 and Qwen 5 are supposed to grow to 5 to 10 trillion parameters. Back then the slides only said "coming soon", with no date, prices or weights.

What we still don't have: an official press release with a date, benchmarks, or confirmation that the 27B model will be released as open weights. So next week is the test of whether Alibaba delivers what developers everywhere are waiting for.

If Qwen4-27B ships open and holds up against paid models, home AI gets a lot stronger.