GLM-4-Plus

GLM-4-Plus is the flagship foundation model Zhipu AI released in August 2024 (debuted August 29 at KDD), positioned as comparable to GPT-4 and Llama 3.1 405B in language understanding, instruction following, and long-text processing. It arrived alongside CogView-3-Plus (text-to-image) and GLM-4V-Plus (image/video understanding), and its consumer debut in Zhipu Qingyan introduced a 'video call' mode — a GPT-4o-style real-time voice-and-vision interaction, the first of its kind from a Chinese lab.

Zhipu, a Tsinghua spinout and one of China's 'AI tiger' companies, later open-sourced GLM-4.5 under MIT, but GLM-4-Plus itself stayed proprietary and API-only. The honest picture: independent English-language benchmarks for the Plus checkpoint are thin — most performance claims come from Zhipu's own reporting — and as of January 2025 the company sits on the US Commerce Department Entity List, which complicates procurement for some organizations. Its strengths are strongest in Chinese-language tasks and long-document reasoning.