Control Plane / Routing

Routing Policies

Define tier preference and fallback chains per role and request type. The data plane (LiteLLM) consumes these policies in later phases; this UI persists the rules in Postgres.

Active Policies

hub

Fallback Chain Architecture

Define the sequential recovery path when primary models experience latency or rate limits.

Create a policy to see the fallback chain.

terminal

Route Simulator

User Context

Request Payload

warning

Degraded Mode

Automatic Throttling

When total gateway latency exceeds 2.5s, drop fallback chains and enforce single-attempt cheap routing.

Error Threshold5% / min

Loading…