GLM-4-Plus is a foundation language model developed by Zhipu AI and officially unveiled in August 2024 at the KDD 2024 conference. It represents a significant iteration of the company's fourth-generation General Language Model (GLM) family, featuring enhancements in language understanding, instruction following, and long-text processing capabilities.
The model is engineered for high-performance reasoning and complex tasks, including mathematical problem-solving and code generation. It supports a context window of 128,000 tokens, utilizing optimized data mixing strategies to improve accuracy in long-context retrieval and reasoning. Zhipu AI has positioned the model as a competitor to other flagship proprietary models, emphasizing its utility in both Chinese and English linguistic contexts.
GLM-4-Plus also provides the foundation for multimodal variants such as GLM-4V-Plus, which incorporates visual and temporal video understanding. While some smaller models in the GLM-4 series are open-sourced, GLM-4-Plus is primarily accessible as a proprietary service through the BigModel open platform API for enterprise and developer applications.