What is GLM-Legal?
GLM-Legal is a domain-specialized artificial intelligence model developed by Z.ai for legal reasoning, document analysis, professional legal research, and agentic workflow execution. Introduced in October 2026, the model is based on the GLM-5.3-Flash architecture and incorporates additional training focused on legal knowledge and practical legal assignments. Its training process combines specialized legal corpora, domain-focused mid-training, and reinforcement learning across simulated legal work environments. GLM-Legal is designed to analyze source documents, identify material facts, interpret applicable legal principles, and develop structured arguments supported by relevant evidence. Contract review capabilities enable the model to evaluate agreement provisions, identify legal and commercial risks, and recommend specific changes to contractual language. Its evidence analysis functionality supports the examination of multiple documents, identification of inconsistent statements, and development of arguments based on referenced source materials. The model can also conduct compliance research involving privacy regulations, cross-border data transfers, marketing requirements, and other jurisdiction-specific legal obligations. GLM-Legal supports multistep agent workflows that combine contextual information retrieval, document analysis, legal reasoning, and generation of formatted work products. Planned integration with AutoClaw enables legal professionals to use these capabilities in document editing workflows, including Microsoft Word files with tracked changes and comments. Z.ai reports a 19.2% all-pass rate on Harvey's Legal Agent Benchmark, representing a 12.5-percentage-point improvement over the underlying base model, alongside competitive results on LegalBench and PLawBench. GLM-Legal is available for beta applications, with Z.ai offering prospective users an opportunity to request API access and explore integration into legal technology.
Screenshots and Video
