Model routing

Organization Settings > Models is the routing engine\u2019s control room: which models are enabled, how Auto resolves, and what happens when a model goes down.

The routing engine

The page opens on a live view of the routing engine itself: enabled models grouped by provider and tier (Premium, Standard, Economy), each with a health indicator and its current position in the Auto resolution chain. This is the same chain the Roles editor prints as a receipt, so what an admin configures here is exactly what a role left on Auto actually gets.

The Models admin page showing the routing engine, enabled models by tier, and shadow/failover configuration
↳The chain every Auto request resolves through, laid out where an admin can actually change it.

Shadow and failover

Beyond simple enable and disable, the page exposes shadow routing (send a fraction of live traffic to a candidate model without switching production over) and failover targets (which model takes over automatically when the primary is unreachable). Both keep a model swap from being an all-or-nothing bet.

Org-wide Complex access - whether the org has switched on higher-tier models for everyone by default - is also managed from this page; per-person and per-group overrides live on the Users and Access Groups pages instead.