Overview
Released on November 16, 2026, Qwen4-Max is Alibaba’s flagship multimodal model for complex professional, creative and research tasks. Alibaba describes it as a sparse mixture-of-experts model with 2.4 trillion total parameters and 71 billion parameters activated per token. (alibabacloud.com)
Capabilities and uses
- Accepts text, images and video and produces text. Supports deep reasoning, web search, function calling and structured output, with OpenAI and Anthropic API compatibility. (alibabacloud.com)
- Intended uses include coding, office work, front-end design, visual analysis and complex agent workflows. Alibaba also describes demonstrations involving 3D production, manufacturing and scientific discovery; these are reported examples, not guarantees of results. (alibabacloud.com)
- The context window is 1,000,000 tokens. Documented maximum output is 131,072 tokens in non-thinking mode or 262,144 tokens shared by thinking and output in thinking mode. (alibabacloud.com)
Limitations and access
Output is text rather than generated images or video. Model Studio documentation lists fine-tuning and model deployment as unsupported. Batch inference support varies by region: it is listed as supported in Beijing but not in Singapore or the listed global regions. Results for consequential tasks should be independently reviewed. (alibabacloud.com)
The model is available through Alibaba’s API and Token Plan, as well as Alibaba applications including QwenWork and QwenStudio. (alibabacloud.com)
Pricing
Prices below are official pay-as-you-go model rates for qwen4-max, checked November 18, 2026. Rates are per 1 million tokens, apply to requests with more than zero and up to 1 million input tokens, and cover both thinking and non-thinking modes. Output billing includes thinking tokens and the answer.
- Alibaba Cloud China site, Beijing: CNY 10 input and CNY 30 output. Beijing Batch File is CNY 5 input and CNY 15 output per 1 million tokens. The China-site Singapore international deployment is CNY 10.938 input and CNY 32.813 output. Official China pricing. (help.aliyun.com)
- Alibaba Cloud international site, Singapore international deployment: USD 1.50 input and USD 4.50 output. The Singapore console lists implicit-cache-hit input at USD 0.075 per 1 million tokens. Official international pricing; Singapore model console. (alibabacloud.com)
- Alibaba Cloud international site, Beijing, Hong Kong, Frankfurt, Virginia and Tokyo: USD 1.413 input and USD 4.24 output per 1 million tokens; deployment scopes vary by region. Official international pricing. (alibabacloud.com)
Regional, cached, batch and subscription prices have different conditions and should not be treated as currency conversions of one package. Tool calls may incur separate charges.