Alibaba Qwen4-Max Preview: 1M Context, Two-Week Trial

Alibaba opened access to Qwen4-Max-Preview on November 11, 2026, marking the first public model from its Qwen4 generation. The multimodal reasoning model ships with a 1-million-token context window and supports text, image and video inputs, but availability is tightly controlled through a two-week preview window that ends November 25, 2026.

Unlike typical model launches, Alibaba released no benchmarks, parameter counts or pricing details, positioning this as a constrained developer trial rather than a production-ready deployment.

What Qwen4-Max-Preview Offers

The model accepts text, images and video as inputs and returns text responses with up to 393,216 output tokens shared between reasoning steps and final answers. Thinking mode is enabled by default, controlled by a reasoning_effort parameter with three settings: low, medium or xhigh, with xhigh as the default configuration.

Qwen Cloud documentation describes Qwen4-Max-Preview as built for coding, office productivity, front-end design and complex agentic workflows. The endpoint ships compatibility with OpenAI Chat Completions, OpenAI Responses and Anthropic Messages APIs, plus support for function calling, web search, built-in tools and structured output in JSON Object or JSON Schema formats.

Token Plan Exclusive Access

Developers can only access Qwen4-Max-Preview through Alibaba's Token Plan subscription service at a dedicated base URL. Pay-as-you-go API access is not supported during the preview period. Users must obtain a Token Plan API key, point requests to https://token-plan.maas.qwencloudapi.com and set the model ID to qwen4-max-preview.

The preview window runs from 10:00 on November 11, 2026 to 24:00 on November 25, 2026 in UTC+8 timezone. Alibaba stated the model will not change during this period and will be permanently retired when the window closes.

Release Date and Retirement Timeline

Officially launched on November 11, 2026, the model operates under a strict 14-day availability constraint. This deliberate scarcity suggests Alibaba is gathering developer feedback and usage telemetry before committing to a full production release of Qwen4-Max or related models in the Qwen4 family.

The retirement date of November 25, 2026 is fixed, giving developers a narrow evaluation window. No extension or migration path to a successor model has been announced.

Missing Documentation and Quality Signals

At launch, Alibaba published no benchmark table, model card, parameter count, license terms or per-token pricing for Qwen4-Max-Preview. This lack of transparency means quality claims about the model remain unverified by third parties and rely solely on the Qwen team's internal assertions.

• No public benchmarks on standard reasoning or coding evaluations
• No downloadable weights or open-source release
• No disclosed training data composition or cutoff date
• No formal model card documenting capabilities or limitations

What This Means

Qwen4-Max-Preview represents Alibaba's first controlled exposure of its next-generation flagship multimodal model. The time-limited, subscription-only rollout with minimal documentation signals a cautious approach, likely designed to stress-test infrastructure and gather real-world performance data before a broader launch. Developers interested in evaluating the model's 1M-token context and reasoning capabilities have until November 25, 2026 to gain access through Token Plan. The lack of benchmarks and pricing leaves open questions about how Qwen4-Max will position against frontier models from OpenAI, Anthropic and Google when it eventually moves to general availability.