Qwen3.8-Max quietly lost its preview label — Alibaba is now serving it as the flagship

Alibaba is serving Qwen3.8-Max on chat.qwen.ai under the plain id `qwen3.8-max`. The preview suffix is gone. It sits first in the app's model list, above Qwen3.7-Plus and Qwen3.7-Max, and Alibaba's own product copy in the page payload calls it "the flagship model of the Qwen3.8 series, delivering state-of-the-art performance across both language and vision modalities."

We want to be exact about what that does and does not establish, because the coverage and the surface currently disagree. Every writeup still describes this model as Qwen3.8-Max-Preview, available through the Alibaba Cloud Token Plan, Qoder and QoderWork — which is what shipped on July 19 at WAIC. What we verified today is narrower and more specific: on Alibaba's own consumer surface, the preview label has come off and the model is presented as the flagship, in the present tense, with a full product description rather than a preview notice.

What we have NOT verified: a public API general availability, published per-token pricing, a model card, benchmark tables, or the open weights that were promised at the preview. DashScope requires a key we do not hold, so we cannot say what the API surface looks like tonight. The 2.4T parameter count and the "second only to Fable 5" positioning remain Alibaba's own claims, uncorroborated by any independent board.

So: this is a surface expansion and a labelling change, confirmed first-party, not a confirmed API GA. That distinction matters more here than usual, because the gap between the two is exactly where a reader would otherwise assume pricing they cannot get.

The reason we are confident at all is that we read it from Alibaba rather than about it. The model list is embedded in the page as structured data, and it is the same class of evidence that told us Kimi K3 was live in Moonshot's app before its API listed anything.