Alibaba Previews Qwen4-Max for Two Weeks, Subscription-Only and With No Independent Scores Yet
Alibaba opened a preview of Qwen4-Max on November 11, the flagship of its next-generation Qwen family, with text, image and video input, reasoning on by default and a 1M-token context window. API access is through its Token Plan subscription only, with no pay-as-you-go option, and Alibaba says the preview model will be retired after November 25. Alibaba has not published benchmark figures in its model documentation, and Artificial Analysis did not list the model when we checked.
Alibaba has opened the first window onto its next Qwen generation, but on narrow terms: a two-week preview, available only to subscribers, with no published benchmark figures and no independent scores yet. For now Qwen4-Max is something developers can try, not something they can yet compare.
What Alibaba released#
Alibaba Cloud's Model Studio documentation describes Qwen4-Max-Preview as "the next-generation flagship multimodal model in the Qwen family," combining text generation, image and video understanding and deep reasoning, aimed at coding, office work, frontend design and complex agentic tasks. It supports function calling, web search and structured outputs, and is compatible with OpenAI's Chat Completions and Responses APIs and Anthropic's Messages API.
| Spec | Value |
|---|---|
| Context length | 1M tokens |
| Maximum output | 393,216 tokens, shared by thinking and the answer |
| Reasoning | On by default; low, medium or xhigh (default) |
| Access | Token Plan subscription only; no pay-as-you-go |
| Preview period | November 11, 10:00 to November 25, 24:00 (UTC+8) |
The documentation says no changes will be made during the preview and that the model "will be retired after the preview ends." TechNode reported that the model is also available for early use in the Qwen app, Qoder, QwenWork and QwenStudio.
One practical change for developers: starting with this model, max_tokens in Chat Completions covers thinking and the final answer together, so a limit sized for the answer alone may truncate output, Alibaba warns.
Claims and checks#
Alibaba's model page carries no benchmark table. When we checked Artificial Analysis's model leaderboard on November 11, it listed Qwen 3.8-series models but no Qwen4 entry. We could not load LMArena's leaderboard. Any comparison with other frontier models therefore has to wait for independent testing. Alibaba's previous flagship also arrived as a preview first: Times of AI covered the Qwen 3.8 Max preview in July.
Limits#
- Alibaba has not said what model, pricing or access terms will follow the preview.
- No vendor or independent benchmark results were available to us; Artificial Analysis showed no Qwen4 listing at our check, and LMArena was unavailable.
- TechNode's report is brief and draws on Alibaba and Cailian Press.