Skip to main content
Router evals are model scorecards. Create one from Evals in the dashboard, then upload a CSV:
model_id and score are required. Higher scores are better. thinking_level and notes are optional. Model IDs must exactly match the provider-prefixed IDs enabled on the router, and each model/level pair must be unique. Import the eval from the router create or edit page. Dari prefers a score matching both model and reasoning level, then falls back to that model’s row with a blank level. Scores inform selection; they do not define a fixed formula.