According to the State Administration for Market Regulation, China has officially rolled out a set of national standards tailored for large artificial intelligence models. This move signifies that China's large model industry has entered a new development phase marked by "scientific rigor and unified standards." As the inaugural national standard specifically targeting general-purpose large models, this set of standards bridges the gap in the technical evaluation system by clearly defining requirements for performance, safety, and service capabilities.

The evaluation tools associated with these standards have obtained accreditation from the China National Accreditation Service for Conformity Assessment (CNAS). So far, these standard tools have undertaken over a thousand evaluation tasks, triggering the use of large models more than 950,000 times. They have precisely pinpointed common problems, such as issues in hallucination control (where AI generates inaccurate or nonsensical information) and content safety. Moreover, they have aided nearly 30 vendors in achieving technological advancements, fostering the creation of a closed loop encompassing "research and development—evaluation—application—upgrading."