Alibaba previewed the Qwen 4 lineup at its Apsara conference on September 22, but stopped short of launching a callable product.
The Qwen 4 overview describes four models still in training: Qwen 4 Max as the flagship, Qwen 4 Flash for throughput and latency, Qwen 4 Plus as the balanced tier, and Qwen 4 27B for local deployment. Alibaba disclosed no model API identifiers, pricing, weights, context limits, public benchmarks or release date.
For builders, this is a roadmap signal rather than a migration trigger. Flash could become the tier to watch for high-volume automation, while the 27B model may suit local workflows. Both need endpoint details, cost data and tool-use testing before teams can plan production adoption.