Skip to content
Rush Commerce
AI & Automation2 min read

Qwen4-Max preview: 1M context, retires November 25

Alibaba's Qwen4-Max preview has a 1M-token context, Token Plan access only, and retires November 25. Test it in your coding tools; don't build a product on it.

Alibaba's next flagship is here, with an expiry date. The Qwen4-Max preview opened on November 11 with a 1-million-token context window and text, image and video input. It's also available only through a subscription, only in one region, and it will be retired on November 25. That's a two-week test drive, not a platform. Here's what the official docs say.

What actually happened

According to Alibaba Cloud's Qwen4-Max-Preview model page:

  • The model. Model ID qwen4-max-preview, described as the next-generation flagship multimodal Qwen model. Context is 1M tokens; max output is 393,216 tokens, shared between thinking and the answer. Thinking is on by default, with reasoning_effort set to low, medium or xhigh.
  • The access. Token Plan only. Pay-as-you-go is not supported. The API accepts OpenAI Chat Completions, OpenAI Responses and Anthropic Messages formats.
  • The window. The preview runs from November 11 to November 25 (UTC+8). Alibaba says the model "will be retired after the preview ends."

The Token Plan Personal Edition page fills in the rest:

  • Region. Singapore only, with global inference that can move data across borders.
  • Price. Plans run $6 to $43 a month for 20,000 to 200,000 credits. New users get Lite at $3 and Essential at $5 until November 30.
  • Discount. Qwen4-Max-Preview calls use credits at as little as 20% of the standard rate, and 8% between 22:00 and 08:00 Beijing time. Alibaba hasn't published the standard rate itself.
  • The catch. The plan is for interactive use in coding and agent tools. Automation scripts and custom backends are not allowed.

Alibaba hasn't released Qwen4 weights, parameter counts or benchmarks.

Why the Qwen4-Max preview matters for your business

You can't ship on it. No pay-as-you-go, no backends, and a retirement date two weeks out. Any feature you build on qwen4-max-preview breaks on November 26.

You can learn from it. For $3 to $6, a developer can point a coding agent at it and see how it handles your real repo. That tells you whether the GA model is worth planning for.

Check the data path first. Singapore-only with global inference means your code may cross borders. Use a sandbox repo, not client code under an NDA.

Key takeaways

  • Qwen4-Max preview launched November 11 with a 1M-token context and multimodal input
  • Token Plan only, Singapore region, no pay-as-you-go
  • Retires November 25; the plan bans automation scripts and custom backends
  • Plans start at $6 a month, or $3 for new users until November 30
  • No weights, parameter count or benchmarks published

New models drop every week. Your product shouldn't depend on any one of them. We build AI systems with a model routing layer and evals on your own data, so you can try a model like Qwen4-Max without betting your app on it. See how we build, or run the numbers on your AI spend.

Sources: Alibaba Cloud: Qwen4-Max-Preview, Alibaba Cloud: Token Plan Personal Edition.

  • #qwen4-max
  • #alibaba
  • #model-pricing
  • #ai-agents
  • #coding-agents
TR

Tommy Rush — Founder, Rush Commerce

Operator turned builder. 15+ years running operations — now shipping the systems businesses run on. More

Get The Rush Report weekly — one email, zero fluff.

Keep reading

AI & Automation3 min read

Notion AI chat transcripts exposed in Fireworks breach

The Fireworks AI breach exposed Notion AI chat transcripts used for Insights in under 1% of workspaces. Map every subprocessor behind your AI features.

Read it
AI & Automation3 min read

n8n Nod: an open-source personal AI agent that asks first

n8n Labs released Nod, an Apache 2.0 personal AI agent for Mac and Windows with its own sandbox and browser. Pilot it on one boring job with a spend cap.

Read it