Qwen3.8 Max Preview

A status page snapshot from July 21, 2026 confirms the exact hosted model ID for the Qwen3.8 preview is qwen3.8-max-preview, offered as a hosted preview through the Qwen Cloud Token Plan rather than a downloadable checkpoint. The Qwen Cloud model table lists it with a 1M-token context window, thinking support, function The Qwen team announced a new preview build live on July 20 with reported broad gains, including a substantial improvement on web frontend work, and described the model as continuing to evolve daily while soliciting failure reports. The page emphasizes that these are vendor performance claims without a public benchmark

A July 20, 2026 article reports that Alibaba previewed Qwen3.8-Max-Preview at the World Artificial Intelligence Conference, citing the 2.4 trillion-parameter figure and multimodal support for images, video, and documents as the first Qwen-family model to cross the trillion-parameter threshold. Alibaba framed the system The article adds independent scrutiny that the preview arrived without supporting technical documentation: no benchmark data, no model card, and no activated-parameter count, and notes that no independent leaderboard has scored Qwen3.8-Max-Preview while Fable 5 holds the top spot on LMArena and the predecessor Qwen3.7-

Economic Times' ETtech wire reports that Alibaba launched Qwen3.8-Max-Preview on July 19, 2026, with 2.4 trillion parameters, positioning it ahead of an open-weight release planned for later. The preview is distributed through Alibaba's Token Plan, Qoder, and QoderWork, giving developers immediate API access to the new The article quotes Alibaba's X post stating that Qwen3.8-Max-Preview is "one of the most powerful models available today, comparable to leading frontier AI models, second only to Fable 5," framing the preview as a competitive response in the ongoing frontier-model race. While the headline associates Fable 5 with Anthro

On July 19, 2026, during the World AI Conference in Shanghai, Alibaba's Qwen team previewed Qwen3.8-Max-Preview, the next flagship in the Qwen family, as reported by MarkTechPost. The Qwen account described the model as "one of the most powerful available today, comparable to leading frontier systems," and Qwen develop The preview build is accessible through Alibaba's Token Plan subscription at 10% of standard pricing, according to the excerpt. Alibaba described Qwen3.8 as having approximately 2.4 trillion parameters, calling it "second only to Fable 5" among systems it tested, though MarkTechPost flags that the benchmark table, mode

Alibaba Group unveiled Qwen3.8-Max-Preview at the World Artificial Intelligence Conference in Shanghai on July 19, 2026, describing it as the most powerful model in the Qwen family and the first Qwen above 1 trillion parameters to natively process images, video, and documents alongside text. The model is sized at 2.4 t Per SiliconANGLE's coverage, Alibaba is positioning Qwen3.8-Max-Preview as trailing only Anthropic's Claude Fable 5 on the frontier, though that ranking is Alibaba's own self-report and was not independently benchmarked in the supplied evidence. The preview availability is tied to the Alibaba Cloud Model Studio ecosyst