Routing, fallback & retry policies
Production resilience across 20+ models — usage/latency/cost routing, fallback chains, and per-org retry, timeout, and cooldown tuning.
- Routing strategies: usage, latency, cost, shuffle, least-busy
- Fallback chains retry on backup models on error or timeout
- Tag-based capability routing — vision, code, long-context
- Every routing decision captured in observability