Skip to content

An editorial team and a community building with AI

← All articles

News · Alibaba / Qwen · 2 min read

Event date: · Published:

Alibaba previews Qwen4-Max with a 1 million token context window

Alibaba Cloud opened a time-boxed preview of qwen4-max-preview on November 11, 2026, with a 1 million token context window, multimodal input and promotional pricing, but no open weights.

Alibaba Cloud opened a time-boxed public preview of qwen4-max-preview on November 11, 2026, offering the first look at the Qwen4 model family through its Qwen4-Max-Preview model documentation. It succeeds Qwen3.8, pairing a very large context window with multimodal input, but without open weights or a permanent availability commitment.

What's in the qwen4-max-preview release

  • Model ID qwen4-max-preview is the first public preview of the Qwen4 family, the successor to Qwen3.8.
  • Input context reaches 1,000,000 tokens, with combined reasoning-plus-answer output capped around 393,216 tokens.
  • Multimodal input covers text, images and video up to 2 hours, 2GB and 64 videos per request; output stays text-only.
  • The preview runs from 10:00 on November 11 to 24:00 on November 25, 2026, Beijing time, after which the model retires.
  • Access requires Alibaba's Token Plan subscription, plus early access inside QwenWork, Qoder, the Qwen App and QwenStudio.
  • Promotional pricing cuts credits by 80%, rising to 88% off during 22:00-08:00 Beijing time off-peak hours.

What it means for AI builders

Qwen already powers much of the open and commercial coding world through tools such as Qwen Code CLI and Qoder, so a 1 million token flagship preview with explicit coding and agentic claims signals where Alibaba is taking its next generation coding model. Alibaba says qwen4-max-preview improves on Qwen3.8 in world knowledge, code generation and sustained multi-turn tool-calling, and defaults to a hybrid thinking mode with function calling and structured output, though it has not published benchmarks or parameter counts.

Availability and next steps

During the preview, access is limited to the Singapore region, and the Token Plan terms prohibit backend or automation use, ruling out unattended pipelines. Alibaba Cloud also posted the launch in its announcement on X. Because no open weights accompany this release, unlike Qwen's usual pattern, and the model retires once the window closes on November 25, 2026, builders should treat qwen4-max-preview as an evaluation window rather than a production-ready option.


Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

Alibaba previews Qwen4-Max with a 1 million token context window | vibecoding.inc