Qwen Chat's own code makes qwen4.0-max-preview the default model for agent mode, before any Qwen 4 release

Qwen Chat's web app now names a Qwen 4 model in its own code. Build 0.3.40 of Qwen Chat's front end, the build chat.qwen.ai was serving when ArtificialWatch read it at 19:52 UTC on November 10, makes 'qwen4.0-max-preview' the default model for agent mode. The previous build, 0.3.39, set 'qwen3.8-max'. @LuminaBench flagged the change at 19:22 UTC.

WHAT IS IN THE CODE

• The agent-mode default model id changes from qwen3.8-max to qwen4.0-max-preview. The code uses that default unless the user picked another model, and falls back to the first model in its list when the id is not available.
• The same file adds an announcement pop-up type called agent_qwen4, with a 'Try in Chat' button.
• Both are in Alibaba's own JavaScript file on its CDN (qwen-chat-fe 0.3.40, main.js), which ArtificialWatch read directly.

WHAT IS NOT LIVE

• Qwen Chat's public models list (read 19:53 UTC) still shows qwen3.8-max as its newest Max model, and OpenRouter lists no Qwen 4 id.
• Alibaba has not announced Qwen 4. Qwen's last public hint was its 'The becoming begins' teaser on November 6.

WHAT WE DO NOT KNOW

When Qwen 4 Max ships, whether the API keeps the 'preview' label, and its price and context length. ArtificialWatch's Qwen 4 Max hunt fires on qwen4.0-max-preview: it sends an early-version alert the moment a catalog we sweep serves that id, and keeps its main alert for the final Qwen 4 Max.