Alibaba's Qwen3.7-Max runs 35 hours autonomously with 1M-token context

Alibaba's Qwen team has launched Qwen3.7-Max, a proprietary large language model built for the 'agent era'—AI systems that can execute tasks autonomously over extended periods. In internal testing, the model achieved approximately 35 hours of continuous autonomous execution, a significant leap for agentic workflows. It boasts a 1-million-token context window, enabling it to process and reason over massive documents, codebases, or multi-step interaction histories without losing track of earlier context. This makes it suitable for tasks like debugging complex software, managing long-form document generation, or orchestrating office workflows that span multiple sessions.

The model's capabilities extend beyond raw context size. It supports native integration with the Anthropic API protocol, allowing developers using frameworks designed for Anthropic's models to swap in Qwen3.7-Max with minimal changes. This cross-framework compatibility lowers the barrier for enterprises already invested in AI agent ecosystems. Qwen3.7-Max excels at coding, debugging, and multi-step office automation—common pain points for knowledge workers. By combining a massive context window with long-duration autonomous operation, Alibaba positions Qwen3.7-Max as a backbone for next-generation virtual assistants that can plan, execute, and iterate on complex projects without human intervention, potentially reshaping how enterprises automate their most time-consuming tasks.