Zhipu AI has launched GLM-4.6, the latest version of its open-source language model. The update brings a larger 200,000-token context window, improved programming abilities, and stronger logical reasoning. According to Zhipu, GLM-4.6 nearly matches Claude Sonnet 4 in real-world tests, winning 48.6 percent of head-to-head comparisons, and is more efficient than the previous GLM-4.5 release. In eight benchmark tests, GLM-4.6 outperformed models like Deepseek-V3.2-Exp, but still trails Claude Sonnet 4.5 in code generation.
GLM-4.6 is available through Z.ai, OpenRouter, HuggingFace, and ModelScope. Technical documentation can be found at docs.z.ai.
