Alibaba launched Qwen 4 Max on Monday, its most powerful model to date and the first flagship built on the Qwen 4 architecture. With 2.4 trillion parameters, a one-million-token context window and native visual understanding, this mixture-of-experts model activates 71 billion parameters at a time. It leads Chinese models on Arena’s agent leaderboard with a preliminary score of 7.46% and ranks sixth globally in web development with 1,750 points. Through QwenCloud, it costs $1.50 per million input tokens and $4.50 per million generated tokens, a 25% reduction from Qwen 3.8 Max. Unlike several earlier Qwen models and the smaller Qwen 4 models announced for future release, its weights are closed and cannot be downloaded, modified or run privately on developers’ own hardware.
Source: Read the original article


