Alibaba's Qwen 3.7 Max runs autonomously for 35 hours with 1,158 tool calls

Alibaba's latest large language model, Qwen 3.7 Max, turned heads in the developer community after completing a 35-hour autonomous task run on May 25, 2026. During this period, the model executed 1,158 uninterrupted tool calls without any human intervention. The test highlights the model's ability to sustain complex, multi-step workflows over extended durations—a challenging benchmark for AI agents. Early reports from overseas developers praise the consistency and reliability of the system, noting that such performance could unlock new possibilities for automated tasks in enterprise environments where long-running processes are common.

The demonstration underscores Alibaba's push into autonomous AI agents, a space increasingly dominated by models like OpenAI's GPT-4o and Google's Gemini 2.0. Qwen 3.7 Max's ability to maintain smooth tool call sequences for hours suggests advancements in memory management and error tolerance. For developers, this means fewer checkpoints and restarts when building automation pipelines. The model's open-source nature (via Alibaba's Qwen family) also allows fine-tuning for specific long-horizon tasks. While real-world deployment will require further validation, this test positions Qwen 3.7 Max as a strong contender for high-reliability automation systems.