Alibaba Releases Qwen4-Max With Lower API Prices

Alibaba released Qwen4-Max with a million-token context window, image and video understanding, and API access. Singapore-region pricing starts at $1.50 per million input tokens and $4.50 per million output tokens. The model’s weights remain closed.

Alibaba Releases Qwen4-Max With Lower API Prices

Alibaba released Qwen4-Max on November 16, opening pay-as-you-go API access to its new flagship model. Singapore-region pricing is $1.50 per million input tokens and $4.50 per million output tokens, according to the QwenCloud model page. Implicit cache hits cost $0.075 per million tokens.

The release follows a subscription-only preview introduced on November 11. The production model uses the identifier qwen4-max and supports text, image, and video input, with text output.

Qwen4-Max uses a mixture-of-experts architecture with 2.4 trillion total parameters and 71 billion active for each token. Alibaba says the active parameter count is 25% lower than Qwen3.8-Max at the same total model size. Its advertised context window holds up to one million tokens.

The model supports tool calling, structured outputs, web search, and context caching. QwenCloud offers OpenAI-compatible and Anthropic-compatible interfaces. Alibaba’s developer documentation describes workflows spanning repository work, document analysis, presentations, and long-running agents.

In one launch demonstration, an agent builds an Unreal Engine game. Other examples involve turning production-line layouts into 3D models and producing financial research with linked spreadsheets and editable presentations. These are Alibaba’s demonstrations, rather than independent evaluations.

Alibaba reports lower task costs; independent tests show slower execution

On its proprietary Work360 benchmark, Alibaba reports 35% lower token use at High reasoning effort and roughly 60% lower average task cost than Qwen3.8-Max. Work360 covers industry tasks across 721 specialized fields, according to the company’s announcement.

Artificial Analysis gives Qwen4-Max an Intelligence Index score of 43, above the median of 41 for its comparable model group. The same evaluation records 32.1 minutes per task, against a group median of 13.3 minutes. Those measurements reflect the benchmark workload and Alibaba’s API, not every coding task.

Qwen4-Max is available through QwenCloud’s API and Token Plan, alongside Alibaba applications including QwenWork and QwenStudio. The flagship remains proprietary. Alibaba has announced smaller open-weight Qwen4 models, but did not provide a firm release date in its launch statement.

Comments (0)