Earlier this week, on December 23, 2025, the Chinese mainland's national supercomputing network unveiled a new scientific-computing intelligent agent in Tianjin Municipality.
By understanding natural-language instructions, the agent can automatically decompose research tasks, schedule compute resources, invoke simulation packages, analyze results, and generate reports. What once took a full day now wraps up in roughly one hour, and the system already supports nearly 100 high-frequency scientific-computing scenarios.
Backed by a rich AI community and repository, the agent draws on more than 120 domain-specific knowledge bases across seven key areas—AI, scientific intelligence, industrial simulation, materials science, and more—lowering the entry barrier for scientific computing and accelerating research productivity.
Key features:
- Automatic task decomposition and resource scheduling
- Simulation package invocation and results analysis
- One-hour turnaround for day-long tasks
- Support for nearly 100 computing scenarios
Qian Depei, an academician of the Chinese Academy of Sciences, said, "Scientific research is now transitioning from computational science to intelligent science. The emergence and evolution of scientific-computing agents integrate fragmented compute power, toolchains and knowledge resources, offering researchers faster and more accessible innovation support."
As global research teams look to push the boundaries of discovery, the Chinese mainland's AI-powered agent marks a significant step toward smarter, more efficient science.
Reference(s):
China launches scientific-computing agent for supercomputing network
cgtn.com
