Overview
Qwen4-Max-Preview is the first model from Alibaba's Qwen4 generation that developers can call. Qwen Cloud's documentation describes it as the next-generation flagship multimodal model in the Qwen family, built for coding, office productivity, front-end design and complex agentic tasks. It accepts text, images and video, returns text, and has a 1-million-token context window with up to 393,216 output tokens shared by its thinking and its final answer.
Thinking is on by default and is set with a reasoning_effort parameter (low, medium or xhigh, default xhigh). The endpoint is compatible with the OpenAI Chat Completions, OpenAI Responses and Anthropic Messages APIs and supports function calling, web search, built-in tools and JSON Object / JSON Schema structured output.
The preview is deliberately narrow. It runs only through Alibaba's Token Plan subscription, at a dedicated base URL, from 11 to 25 November 2026 (UTC+8), and Alibaba says the model will not change during that window and will be retired afterwards. At launch Alibaba had published no benchmark table, model card, parameter count, license or per-token price, so quality claims about Qwen4-Max remain the Qwen team's own.
| Released | 2026-11-11 |
|---|---|
| License | Proprietary (preview); no license or open-weight release announced. |
| Weights | API only |
| Context | 1M tokens (max output 393,216 tokens, shared by thinking and the answer) |
| Modalities | Text, Image, Video |
| Status | Preview. Available through Alibaba's Token Plan only (no pay-as-you-go) from 10:00 on 11 November 2026 to 24:00 on 25 November 2026 (UTC+8); the model is frozen during the window and retired when it ends. |
Strengths
- 1M-token context with a 393,216-token output budget shared by thinking and the answer
- Text, image and video input, aimed at screenshot analysis, document understanding and video analysis
- Adjustable reasoning effort (low, medium, xhigh) with thinking on by default
- One endpoint that speaks OpenAI Chat Completions, OpenAI Responses and Anthropic Messages, so coding agents can use it unchanged
Best for
- Trying Qwen's next flagship inside coding agents that support Token Plan, such as Qwen Code, Cline, Claude Code, Cursor, OpenCode and Codex
- Long-context coding, office and front-end design tasks
- Multimodal work over screenshots, documents and video
- Note: the preview is retired after 25 November 2026 and has no published benchmarks — not a base for production work
How to access
| Provider | Model ID |
|---|---|
| Alibaba Token Plan ↗ | qwen4-max-preview |
Qwen-Max — every version
The full lineage of the Qwen-Max line, newest first. Every version has its own page — click any to compare specs, benchmarks and pricing.
| Version | Released | Context | License |
|---|---|---|---|
| Qwen4-Max-Preview | 2026-11-11 | 1M | Proprietary |
| Qwen3.8-Max-0902current | 2026-09-02 | 1M | Proprietary |
| Qwen3.8-Max | 2026-08-03 | 1M | Proprietary (weights published as Qwen3.8-2.4T-A95B) |
| Qwen3.8-Max-Preview | 2026-07-19 | — | Proprietary |
| Qwen3.7-Max | 2026-05 | — | Proprietary |
| Qwen3-Max | 2025-09 | — | Proprietary |
| Qwen2.5-Max | 2025-01-29 | — | Proprietary |
FAQ
How do I get access to Qwen4-Max-Preview?
Qwen4-Max-Preview is served only through Alibaba's Token Plan; pay-as-you-go API access is not supported. Take a dedicated Token Plan API key, use the base URL https://token-plan.maas.qwencloudapi.com and set the model ID to qwen4-max-preview.
When does the Qwen4-Max preview end?
The preview runs from 10:00 on 11 November 2026 to 24:00 on 25 November 2026 (UTC+8). Alibaba says the model will not change during that window and will be retired when it ends.
Does Qwen4-Max have benchmarks or open weights?
Not at launch. Alibaba published no benchmark table, model card, parameter count, license or downloadable weights for Qwen4-Max-Preview.