Qwen4-Max uses a mixture-of-experts architecture with 2.4 trillion parameters. The model supports a context window of up to one million tokens. It also accepts images directly as input.
According to Alibaba, Qwen4-Max uses fewer tokens than Qwen3.8-Max on comparable tasks. The model is also available through Qwen Studio, QwenWork and Qoder.
Qwen4-Max is a closed model, while the company said smaller-scale, open-weight versions of the Qwen4 series would soon be launched for global developers. Alibaba plans to release smaller Qwen4 models as open source.