In addition, Kunlun Large Model uses “digital employees” to support management and services. 23 professional and efficient “digital employees”, such as “Digital Accounting Audit Specialists” and “Digital Cost Engineers”, have taken up their posts, comprehensively enhancing business efficiency and effectiveness.

Currently, the Kunlun Large Model AI Middle Platform has embedded three major production lines: corpus processing, model training, and application deployment. It has achieved unified management of multiple commercial large model training and inference toolchains and supports the private deployment of mainstream open-source large models. A full-stack domestic intelligent computing power environment has been established, with all computing power resources incorporated into the AI middle platform management. At peak times, the computing power usage scale reached 1,950P.

China National Petroleum Corporation has established a three-level dataset management mechanism at the industry, professional and scenario levels. It has formulated 73 data collection and annotation norms as needed, developed dedicated data processing tools, and improved the efficiency and quality of data processing. The scale of the dataset has reached 500TB, with 1 million industry knowledge Q&A pairs and 250,000 industry reasoning Q&A pairs.