China’s Zhipu AI launches new major model GLM-5 in challenge to its rivals

South China Morning PostCenter-RightEN 1 min read 100% complete by Vincent ChowFebruary 11, 2026 at 10:22 PM

AI Summary

short article 1 min

Zhipu AI, a Chinese AI company, has released its new flagship model, GLM-5, to compete with other Chinese tech firms. GLM-5 boasts improved coding capabilities and a shift towards AI-automated coding at a larger scale. According to Zhipu AI's internal tests, GLM-5 surpasses Google DeepMind’s Gemini 3 Pro in coding and agentic performance for open models, but lags behind Anthropic’s Claude in coding benchmarks. The new model doubles the size of its predecessor with 744 billion parameters and increased training data, utilizing DeepSeek Sparse Attention architecture for efficiency. GLM-5 is currently available on Zhipu’s website and has been open-sourced on GitHub and Hugging Face.

Key Entities & Roles

Key Player Opposition Context

GitHub 30% Hangzhou 30% GLM-4.7 40% Nvidia 40% Claude 50% Gemini 3 Pro 50% DeepSeek 50% Anthropic 60% Google DeepMind 60% China ×270% GLM-5 ×480% Zhipu AI (Z.ai) ×490%

Keywords

artificial intelligence 100% zhipu ai 90% glm-5 90% large language model 80% coding capabilities 70% model performance 60% agentic engineering 60% deepseek sparse attention 50% open-source 50% training data 40%