MyPrivateClaw

GLM-5: Zhipu AI's 744B Open-Weight Model Targets Long-Horizon Agentic Tasks | Local AI

Z.ai (formerly Zhipu AI) has released GLM 5, a 744B parameter mixture of experts model with 40B active parameters, claiming top open source rankings on agentic…

Published on MyPrivateClaw

Apr 8, 2026, 8:58 AM UTC

Coverage date

Apr 8, 2026

Last updated

Apr 8, 2026, 12:59 PM UTC

News summary

A New Contender in Open Weight Agentic AI Z.ai (formerly Zhipu AI) has released GLM 5, a 744 billion parameter mixture of experts (MoE) language model targeting complex systems engineering and long horizon agentic tasks. The model scales significantly from its predecessor GLM 4.5, growing from 355B parameters (32B active) to 744B parameters (40B active), while increasing pre training data from 23T to 28.5T tokens. GLM 5 is fully open weight, with model weights available on HuggingFace under the zai org organization, making it one of the largest openly available models as of April 2026. Architecture and Training Innovations GLM 5 integrates DeepSeek Sparse Attention (DSA), which significantly reduces deployment cost while preserving long context capacity — a critical consideration for agentic workloads that require maintaining context across extended task horizons. On the training side,…