Score checks over N trials of one task with a real model. Spends money.
⚠️ paid_ overstates slightly: with an injected evalDriver this drives whatever you supply. The DEFAULT path bills.
paid_
evalDriver
Score a check vocabulary across trials against the real claude CLI.
claude
Score checks over N trials of one task with a real model. Spends money.
⚠️
paid_overstates slightly: with an injectedevalDriverthis drives whatever you supply. The DEFAULT path bills.