Qwen4-Max
Alibaba's flagship proprietary model for its Qwen API and cloud platform
Qwen4-Max is Alibaba’s top-tier proprietary model, sitting above the open-weight Qwen3 family the company keeps releasing on Hugging Face. Unlike those, Qwen4-Max is closed and only reachable through Alibaba Cloud’s API, Qwen Chat, or partner platforms, priced at $1.20 per million input tokens and $4.80 per million output tokens. It takes text and image input, carries a 1 million token context window, and Alibaba has been positioning it directly against GPT-5.6 and Gemini 3.5 Pro in its own benchmark comparisons.
The split between open Qwen3 models and closed Qwen4-Max mirrors a pattern other labs have settled into: give away capable mid-size weights to build developer mindshare while reserving the largest, most expensive-to-train model for a paid API. Alibaba has leaned on this approach to expand Qwen’s footprint well beyond China, particularly with developers on Hugging Face who fine-tune the open releases but still reach for the Max tier in production when they need the extra reasoning headroom.