Route AI requests across hosted models, open-source inference, and private endpoints with automatic fallback, cost-aware routing, usage controls, and observability.
Any model · Any provider · Any endpoint · One router
One API. Hosted, open-source, or private endpoints.
Cost, latency, quality, region, or business rules.
Stay online when providers degrade or rate-limit.
Budgets, usage limits, and team-level policies.
Latency, cost, tokens, errors, and routing decisions.
Private clusters, self-hosted models, or any endpoint.
Start with one provider or connect them all. One stable API. Every provider competes for every request.