Timeline

Alibaba previews Qwen4-Max

A one-million-token-context model, sold only through Alibaba Cloud's Token Plan subscription until 25 November, with no benchmarks, model card or weights published alongside it.

  • Models & capabilities
  • Notable

Alibaba opened a preview of qwen4-max-preview, a model named for the Qwen 4 generation, which its Qwen team had said in September was in training with no release date.

According to Alibaba Cloud’s documentation page, the model has a context window of 1 million tokens and a maximum output of 393,216 tokens, shared between its reasoning and its final answer. It accepts image and video input, supports tool calling and structured outputs, and runs with thinking switched on by default. It can be called through OpenAI-compatible Chat Completions and Responses endpoints, or the Anthropic Messages format.

Access is restricted. The page says the model is “available through Token Plan only”, Alibaba’s personal and team subscriptions, and that pay-as-you-go access is not supported. The preview runs from 10:00 on 11 November to 24:00 on 25 November, Beijing time, with no changes during that period; the model is then to be retired.

The page publishes no benchmark results, no model card, no pricing and no statement on open weights. Alibaba’s earlier Qwen models have often been released as open weights, but nothing on the page says whether Qwen 4 will be. Third-party headlines claiming Qwen4 beats rival models cite no Alibaba source.

The preview gives developers a chance to test the flagship tier ahead of a family release that Alibaba has not scheduled. Whether it returns as a general product, as open weights, or both is still undeclared.