
Tech • IA • Crypto
Alibaba has unveiled Qwen 3.7 Max, a high-end AI model optimized for long-running multi-agent systems, claiming major gains in cost, scalability, and autonomy.
Alibaba introduced Qwen 3.7 Max, a new AI model positioned for complex agent-based workflows. It is described as significantly cheaper—up to four times less expensive—than competing systems from OpenAI and Anthropic, while aiming to match or exceed their capabilities in specific use cases.
The model is designed to coordinate large numbers of autonomous agents working toward a shared task. It can reportedly sustain operations for up to 35 hours, managing dozens of agents simultaneously without system failure, highlighting a shift toward persistent, collaborative AI systems.
A key feature is support for up to 1,000 MCP connections, enabling integration with numerous tools and services. This allows agents to dynamically interact with external systems and maintain continuity, even when individual agents fail and are replaced mid-process.
While not surpassing top Western models across all benchmarks, Qwen 3.7 Max stands out in long-duration, tool-integrated automation. Its strengths lie in reliability, endurance, and coordination rather than raw general intelligence performance.
Alongside the model, Alibaba is advancing fully domestic infrastructure, including AI chips and data centers. This reflects a broader push by Chinese firms to reduce reliance on NVIDIA and other foreign technologies amid ongoing trade restrictions.
The release signals growing competition in agentic AI systems, particularly in industrial and enterprise automation. China’s progress suggests a move toward vertically integrated AI ecosystems, combining models, hardware, and infrastructure.
Alibaba’s latest release underscores a strategic push into scalable, autonomous AI systems, potentially reshaping competition in enterprise automation and global AI infrastructure.