Alibaba's Qwen3.7-Max Runs 35 Hours Autonomously

Alibaba's Qwen Team has released Qwen3.7-Max, an agentic AI model capable of 35 hours of continuous autonomous execution. The model supports external tools like Anthropic's Claude Code, marking a shift toward multi-day task planning and execution. Unlike previous Qwen releases, this model is proprietary rather than open source — a departure that follows the exit of several key Qwen Team leaders. Analysts see the closed approach as financially motivated, given the high cost of training powerful AI models.