Alibaba Cloud CTO Feifei Li told an Apsara Conference audience on October 29 that the Qwen4 family would launch "next week," putting the company’s next model push on a tight clock and sending developers looking for the release details that usually follow. By November 3, though, the basic public paperwork was still absent: no release post, no model weights, no model card, no license, no benchmark results and no official price had surfaced in the records reviewed.
That gap is why Qwen4 is being searched now. Alibaba had already said on September 22, 2026 that Qwen 4 was in training, and materials tied to the October 29 event showed four names in the lineup: Qwen4-Max, Qwen4-Plus, Qwen4-Flash and Qwen4-27B. The company had been building toward this moment in stages, first with an architecture preview, then with training confirmation, then with the lineup reveal, and finally with a promise that the launch was days away.
The weight of that promise is not just in timing. If a model family is about to go public, developers need to know what can actually be downloaded, what is covered by the license, how it performs, and what it costs. None of that had been made public by November 3, and that leaves the launch looking more like an announcement of intent than a finished release.
The open question is sharper on Qwen4-Max than on the rest of the family. Alibaba has not clearly confirmed that every Qwen 4 variant will be open-weight, and that matters because a single exception would change how much of the lineup outside developers can inspect, run or adapt. Alibaba had scheduled Qwen3.8-Flash-Next and an FP8 version for August 26, 2026 at 11 p.m. Beijing time, and describes Qwen3.8-Flash-Next as a preview of the architecture intended for Qwen 4, which makes the missing release materials even more notable.
For now, the story is not that Qwen4 failed to arrive. It is that Alibaba has spent weeks laying out the path to the launch, then left the public still waiting for the proof that the family is actually ready.
