What it is
GLM-5 is a family of large language models from Z.ai, including GLM-5.1, 5.2, 5.3 and a smaller 5.3-Flash, published as open weights under Apache-2.0. The flagship models use a 744B-parameter mixture-of-experts design with 40B active parameters, sparse attention, and up to a 1M-token context. Developers and researchers who run their own inference can download FP8 or BF16 checkpoints to power coding agents, terminal automation, and multi-step tool-calling workflows.