Run an A/B eval across arms with a real model. Spends money.
⚠️ The paid_ prefix overstates slightly: pass your own evalDriver and no model of ours is called. The DEFAULT path bills.
paid_
evalDriver
Run an A/B eval across arms with a real model. Spends money.
⚠️ The
paid_prefix overstates slightly: pass your ownevalDriverand no model of ours is called. The DEFAULT path bills.