Today for AI
HOT RADAR
agentHEAT 18.6°

AI CLUSTERED EVENT · 10/13/2026

Zhipu AI Launches GLM-Legal: An Agentic Model Achieving 19.2% Pass Rate on Harvey LAB for End-to-End Legal Workflows

1 reports archived1 independent sourcesupdated 10/13/2026, 00:11:45
Synthesis & Latest Updates
1 Sources Cross-Validated

Zhipu AI has released GLM-Legal, a specialized large language model designed for professional legal collaboration that bridges the gap between fragmented source documents and final work products. Evaluated across complex task execution, legal reasoning, and real-world workflows, the model achieved a 19.2% all-pass rate on the Harvey LAB benchmark while matching frontier closed-source models in domain performance. GLM-Legal focuses on maintaining rigorous domain accuracy throughout the execution lifecycle, enabling agents to extract core issues, build defensible arguments, and format findings to exact specifications.

LATEST/Zhipu AI has released GLM-Legal, a specialized large language model designed for professional legal collaboration that bridges the gap between fragmented source documents and final work products. Evaluated across complex task execution, legal reasoning, and real-world workflows, the model achieved a 19.2% all-pass rate on the Harvey LAB benchmark while matching frontier closed-source models in domain performance. GLM-Legal focuses on maintaining rigorous domain accuracy throughout the execution lifecycle, enabling agents to extract core issues, build defensible arguments, and format findings to exact specifications.

HEAT TRENDHourly heat curve

4 fully observed hours
Now
18.6
Peak
20.101:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

TIMELINECoverage timeline

Total 1 reports · Latest first
  1. Hacker News AIT2·68 pts
    • GLM-Legal optimizes the complete chain from facts and doctrine to argumentation and documentation, emphasizing domain rigor in high-stakes environments.
    • Achieved a 19.2% all-pass rate on the Harvey LAB benchmark, demonstrating capability comparable to frontier closed-source models in handling complex, multi-constraint legal tasks.
    • Core capabilities include extracting key issues from fragmented files, constructing defensible legal arguments, and formatting outputs strictly according to task specifications.
Zhipu AI Launches GLM-Legal: An Agentic Model Achieving 19.2% Pass Rate on Harvey LAB for End-to-End Legal Workflows | Today for AI