China released Qiusuo 4.0 on Tuesday, October 13, the country's first artificial intelligence (AI) evaluation benchmark covering the entire industry chain. It spans computing power, model algorithms, intelligent products, industrial applications and security governance.

The system provides chipmakers, model developers and other manufacturers with unified testing procedures and evaluation criteria, helping make results more comparable across different environments.

It also incorporates real-world business workloads in sectors such as power, healthcare and steel, addressing a longstanding challenge in evaluating AI technologies.

Qiusuo has completed multiple rounds of pilot evaluations and validated products in areas including finance, power and petrochemicals. Its lightweight evaluation tools for computing power, large models and intelligent agents have also been open-sourced, enabling wider adoption and hardware-software compatibility testing.

The benchmark comprises 16 evaluation tools across four areas: foundational software and hardware, models and intelligent systems, security governance, and industry implementation. It covers applications ranging from intelligent computing infrastructure to embodied intelligence and AI security.