Qwen4-Max: New Flagship Neural Network with Million-Token Context
Chinese company Qwen has released the flagship language model Qwen4-Max, which has become the most powerful solution in its class. The new neural network supports a context window of up to one million tokens.
In the Agents' Last Exam coding test, the model achieved a score of 46.7%, surpassing competitors Opus 5.5 and Kimi K3.1 Pro. This is the best result among all solutions available on the market.
Developers paid special attention to resource optimization. The model consumes 50% fewer tokens, and the cost of executing tasks has decreased by 62% compared to the previous version.
Qwen4-Max demonstrates high efficiency in game development thanks to easy integration with Unity and Unreal Engine. Access to the model is available via the Qwen app and API.











