Z.ai disclosed GLM-Legal's training, benchmarks and application-only beta without public weights or a generally available API.
Z.ai opens a limited beta for the GLM-Legal model
2026-10-12Verified announcementWATCH
Z.ai announced GLM-Legal, a GLM-5.3-Flash-based model trained for legal analysis and document workflows. The company reports a 19.2% all-pass rate on Harvey LAB, with Harvey LAB and LegalBench results evaluated by Vals. Access is a limited beta through an application form, and AutoClaw support is described as coming soon; public weights and a generally callable API are not available.
Availability: application-only beta; AutoClaw integration coming soon; no public weights or generally available API
What this means for you
For legal-technology teams, an independently administered benchmark is stronger than a vendor-only score, but limited beta access prevents a normal engineering comparison and legal outputs remain high risk. The useful next signal is reproducible access plus citation, jurisdiction and document-editing error evidence.
Evidence & related material
Technical details & scores
| Base Model | GLM-5.3-Flash |
|---|---|
| Training | specialized mid-training and reinforcement learning across legal environments |
| Vendor Result | Harvey LAB 19.2% all-pass, 12.5 percentage points above base |
| Independent Role | Z.ai states Vals evaluated Harvey LAB and LegalBench |
| Availability | application-only beta; AutoClaw integration coming soon; no public weights or generally available API |
| Time Basis | official page gives 2026-10-12 without time of day |
Technical impact 3
Engineering value 3
Adoption signal 1
Maturity 2
Verification cost 4
Risk 5