Alibaba's Qwen3.7-Max runs 35h autonomously with 1M-token context
Qwen3.7-Max sustains 1,000+ tool calls over 35 hours without human intervention.
At the Alibaba Cloud Summit on May 20, 2026, the Qwen team formally introduced Qwen3.7-Max, their new flagship AI agent model. This proprietary, closed-weight model is purpose-built for agent-centric workloads—meaning it's optimized for autonomous, multi-step tasks rather than simple question-answering. It supports a massive 1M-token context window, enabling it to process entire codebases, long documents, or complex workflows in a single inference pass.
Internally, Qwen3.7-Max demonstrated sustained autonomous execution for up to 35 hours with over 1,000 tool calls—a strong indicator of reliability for long-horizon agents. The model shows particular strength in coding, office automation, and scenarios requiring extended reasoning without human hand-holding. For enterprises, this translates to AI agents that can manage entire business processes, from software development cycles to back-office automation, reducing the need for constant human oversight while maintaining task coherence over long durations.
- 1M-token context window allows processing of massive documents or codebases in one go.
- Sustained autonomous execution of 35 hours with over 1,000 tool calls in internal tests.
- Optimized for agent-centric workloads: coding, office automation, and long-horizon tasks.
Why It Matters
Enterprises can now deploy AI agents that run complex, multi-day workflows autonomously with minimal human intervention.