The newest iteration of the Qwen artificial intelligence series is now available through an API designed for professional workflows in sectors including manufacturing, finance, and scientific discovery. The model, Qwen4-Max, uses a mixture-of-experts architecture and features native vision and a 1M token context window, which the company says reduced the average cost per task by over 60% in internal tests compared to the previous Qwen3.8-Max.
The model ranks in the top five on both the Spreadsheet and Python-PPTX arena leaderboards, which puts it in the same performance bracket as current flagship models from OpenAI and Anthropic. Access is live on QwenCloud with pricing starting at $1.5 per million input tokens and $4.5 per million output tokens, while Alibaba noted that open weights for smaller Qwen4 models will be released in the future.
