[The First Full-Stack Domestic Lightweight Intelligent Agent Large Model Officially Released] According to a report by Jinse Finance, on September 17, China Telecom Artificial Intelligence Technology Co., Ltd. officially released the next-generation Xingchen Large Model Xing4.0-29B-A4B. Designed for the era of intelligent agents, this lightweight code intelligent agent large model focuses on capabilities such as complex task understanding, task planning, tool utilization, and autonomous execution. It supports long-context processing and can be widely applied in scenarios such as code development, data analysis, office automation, and life services. It is reported that the model adopts the MOE architecture, with a total parameter count of 29B and only 4B active parameters. It is the first domestic large model with over 10 billion parameters to be trained using domestic computing power and frameworks, deeply optimized for complex engineering tasks.