Z.ai, a Beijing-based enterprise focusing on artificial intelligence (AI), recently released its next-generation flagship large language model named "GLM-5.2". Built for long-horizon tasks, the model supports an industrially viable context window of up to 1 million tokens. In international comprehensive tests, GLM-5.2 ranks among the top global open-source models, achieving internationally leading performance in core capabilities such as coding and long-horizon tasks.
Z.ai was founded based on the research achievements from Knowledge Engineering Group (KEG) of Tsinghua University's Department of Computer Science and Technology. It is the world's first publicly listed company centered on artificial general intelligence (AGI) foundation models.
The newly released GLM-5.2, through innovation in model architecture and engineering optimization, effectively reduces the computational costs of both training and inference, while demonstrating strong generalization capabilities in areas such as multi-agent collaboration and tool use. The model is currently being deployed across multiple fields, including government services, finance, manufacturing, and healthcare, providing new foundational support for the transition of AI's role from technical breakthroughs to industrial tools.
Industry insiders noted that Z.ai has not adopted the "brute-force scaling" approach of investing in a massive amount of computing power. Instead, it leverages its local talent pool, proprietary technological reserves, engineering optimization expertise, and industrial base, continuously improving model training and inference efficiency through algorithmic architecture innovation and joint hardware-software optimization.
GLM-5.2 has been released as an open-source model under the permissive MIT License, with model weights freely available for download, deployment, and commercial use without geographic restrictions.
With their open models' capabilities, relevant Chinese enterprises are making their own innovation achievements available to global developers, giving research institutions, enterprises, and individual developers more options concerning technological exploration and application innovation.
(Source: Beijing Fabu)
