On December 22, Beijing Zhipu Hua Zhang Technology Co., Ltd. officially released and open-sourced its new generation flagship large language model GLM-4.7, once again causing a stir in the global large model competition, offering developers and researchers new choices. The model has demonstrated outstanding performance in multiple international mainstream public benchmark tests, especially excelling in the coding field, with comprehensive performance surpassing GPT-5.2. In the authoritative coding evaluation platform Code Arena, where millions of users participated in blind testing, it ranked first among both open-source models and domestic models.

QQ20251223-085257.png

GLM-4.7 is deeply optimized for programming scenarios, significantly enhancing code generation, long-term task planning, and external tool collaboration capabilities. Official data shows that in mainstream programming benchmark tests such as LiveCodeBench and SWE-bench, its code generation quality and problem-solving capabilities rank first among open-source models. In mathematical reasoning, it reaches the highest level among open-source models on the AIME 2025 competition benchmark. In intelligent agent task execution, by improving task decomposition and tool calling mechanisms, it significantly enhances autonomous decision-making efficiency in complex scenarios.

Furthermore, GLM-4.7 continues to leverage Zhipu's technical advantages in long context processing and multi-round dialogue consistency, supporting efficient processing of 128K-length inputs, demonstrating high stability and low hallucination rate in real programming tasks. The model is compatible with mainstream inference frameworks such as vLLM and SGLang, facilitating local deployment and enterprise-level integration, and lowering the application threshold.

The release and open-sourcing of GLM-4.7 highlight Zhipu's technical strength and innovative determination in the field of large models, providing global developers with powerful and easy-to-use tools, which are expected to drive innovation and development in multiple fields such as coding and AI research, setting a new benchmark in the competition of open-source large model technologies and leading the industry to new heights.