Benchmark

Evaluation suites for England & Wales legal AI.

A growing catalogue of suites — citation, redaction, real work tasks, and more. Select a suite for status and methodology. When a suite freezes, published results appear here as graphs.

Suites

Citation integrity

Citation resolution suite

Internalv0.3

Internal England & Wales gold-citation replay. Checks whether named authorities resolve and whether the right authority is cited for the claimed point. Public methodology and manifests ship when this suite freezes — v0.3 is not a published leaderboard.

Citation integrity in evaluation (TR-2026-01)

Results

  • Citation resolution—
  • Authority fit—
  • Quote fidelity—

Comparable scores publish when this suite is frozen.

This page lists evaluation suites as we define them. Comparable scores and manifests publish only when a suite is frozen. Internal and planned entries are orientation — not leaderboards.