Z.ai
Z.ai lists a legal GLM at $0.37 but sells it by application only
Z.ai's documentation lists GLM-5.3-FlashX-Legal, a legal-work model priced like FlashX, with API access limited to selected enterprise customers.
Z.ai has a legal-work version of its GLM-5.3-Flash model with a public price, but only selected enterprise customers can get API access, and only by applying.
The model is GLM-5.3-FlashX-Legal, also called GLM-Legal, with the model code glm-5.3-flashx-legal. Z.ai’s documentation describes it as “optimized for professional legal workflows, with a focus on document understanding, legal reasoning, and task execution,” covering legal research, contract review and legal document drafting. The documentation is undated; the page was read Oct. 28.
Access
Z.ai limits API access to selected enterprise customers, who apply through a consultation form on z.ai.
How It Was Built
GLM-Legal is built on the GLM-5.3-Flash family. Z.ai said it went through legal-domain mid-training and post-training, including reinforcement learning in tens of thousands of task environments modeled on real legal workflows. GLM-5.3-Flash has 320 billion total parameters with 18 billion activated and a 1 million-token context window. Z.ai calls it the first native multimodal model in the GLM-5 series.
On Harvey LAB, a benchmark for legal agents, Z.ai said the model’s legal capabilities reach an all-criteria pass rate of 19.2%, 12.5 percentage points above the base model. Z.ai claims the best result among open-source models on that evaluation. The figures are Z.ai’s own, and the documentation does not give an evaluation date.
Z.ai recommends running the model in an agent environment connected to MCP services that supply laws, cases and company information, plus tools for creating and revising Word documents with tracked changes and comments. It also points to AutoClaw Legal Assistant, a workspace with plugins for Word and WPS.
Pricing
Z.ai’s pricing page, as of Oct. 28, lists GLM-5.3-FlashX-Legal at $0.37 per million input tokens, $0.075 per million cached input tokens and $1.25 per million output tokens. That matches the listed price of GLM-5.3-FlashX, the faster tier covered in an earlier article. GLM-5.3-Flash costs $0.15 for input and $0.50 for output, and GLM-5.3 costs $1.40 and $4.40.
Z.ai’s release notes, as of Oct. 28, list GLM-5.3-Flash (Aug. 26, 2026) and GLM-5.3 (Aug. 18, 2026) but carry no entry for the legal model.
Our View
Z.ai lists a public price, yet a law firm or in-house legal team cannot buy the model on it. Their developers have to apply and be selected before they can run a contract through it. Any comparison on their own documents waits on Z.ai’s approval.
The only performance figure also comes from Z.ai. A 19.2% pass rate on Harvey LAB, by Z.ai’s account, shows how the model did on that benchmark. It says little about how the model handles a firm’s contracts. Until a team is selected and tests its own documents, the price and the benchmark are the only things it can weigh. Both come from the vendor. The model pays off only if Z.ai selects the team and it holds up on the team’s own contracts.