Loading scenario evidence…

Each row keeps its original comparison boundary. Evidence from different releases, harnesses, providers, tiers, and scenario versions is shown together for inspection, never silently pooled into a leaderboard.

0published cohorts

0model lanes observed

0/0recorded repairs

n/aknown evidence spend

Published cohorts

Release	Version	Boundary	Harness	Models	Repairs

Model evidence

Cohort	Model	Repairs	Attempts	Pass rate	95% CI	Median	Known cost	Failures

Repair failures

Unavailable trials

← Scenario guide

This profile is generated from benchmark-catalog.json.