LegalAI

Workspace

  • Dashboard
  • Upload
  • Documents
  • AI Assistant

Analysis

  • Review Queue
  • Knowledge Graph
  • Evaluation

Platform

  • Model Router
  • Analytics
  • Admin

Resources

  • Documentation
  • Architecture
  • Research
  • About
  • Settings
  • Back to website
GitHub
Sign in

Evaluation

The graded eval harness behind the routing policy. A task only cuts over to a self-hosted default when its candidate meets or beats the baseline.

Eval runs recorded

0

Cutover gates passed

—

candidate ≥ baseline × ratio

Tasks evaluated

0

Most recent run per task & provider