Routing
Tiers & overrides
Tune auto-mode for your cost and quality bar at org and team level.
Auto-mode is configurable so routing matches your cost and quality preferences.
Org-level overrides
force_tier— pin every auto request to a single tier.max_tier— cap the highest tier auto-mode may select.cost_weight— bias routing toward cost or quality. Higher values prefer economical models; lower values allow powerful ones.simpleMax/complexMin— adjust the score thresholds between tiers.
Team-level overrides
Teams can carry their own auto-mode override on top of org defaults, so a latency-sensitive team can cap tiers while a research team allows the frontier.
Configure these in the dashboard under Settings → Auto-mode, or manage them through the tenant auto-mode API.
Model-to-tier mapping is maintained server-side and evolves as new models ship. Auto-mode applies your overrides on top of the current mapping.