Routing Policies

Routing Policies#

Routing policies control how the Gateway selects a provider for each request. Policies are evaluated in order of priority.

Policy Types#

• cost-optimized: select the cheapest provider that meets latency requirements
• latency-optimized: select the fastest provider
• round-robin: distribute requests evenly across providers
• failover: use a primary provider with automatic fallback

Example Policy#

policies:
  - name: production-default
    type: failover
    primary: openai
    fallback:
      - anthropic
      - google
    healthCheck:
      interval: 30s
      timeout: 5s