ALIBABA'S QWEN3.7-MAX RUNS 35 HOURS TO OPTIMIZE CHIP CODE

Alibaba's Qwen team has unveiled Qwen3.7-Max, a new AI model engineered specifically for long-running autonomous agent tasks. The demonstration of a 35-hour autonomous run optimizing code for the company's custom chip marks a notable capability in extended task execution. The model matches Claude Opus 4.6 on standard benchmarks while outperforming Chinese competitors including DeepSeek V4 Pro and Kimi K2.6. This positions Qwen3.7-Max as a competitive entry in the high-end AI model market. Beyond code optimization, Alibaba's team demonstrated the model's versatility by deploying it to control a four-legged robot, showcasing its ability to handle real-world physical tasks through extended autonomous operation. The release of Qwen3.7-Max reflects ongoing competition in the AI sector around capability benchmarks and practical applications. The model's design focus on long-duration autonomous tasks addresses a specific operational need where traditional models may face limitations in sustained reasoning and decision-making. The 35-hour autonomous run represents a significant operational window for AI-driven optimization work. Extended autonomous operation reduces the need for human intervention in iterative processes, potentially accelerating development cycles for hardware and software projects. Alibaba's emphasis on autonomous agent capabilities comes as multiple AI labs pursue increasingly capable models. The competitive landscape includes ongoing developments from U.S.-based firms and other Chinese AI companies, each targeting different performance metrics and use cases. The model's performance on established benchmarks provides quantifiable comparison points, though real-world deployment effectiveness depends on specific task requirements and integration factors.