# Embeddings API Routing
#
# Routes OpenAI Embeddings API requests to a dedicated Embeddings API
# backend. The router uses segment-boundary prefix matching, so
# /v1/embeddings and all of its subresources are routed to the
# embeddings-api cluster, while paths such as /v1/embeddingsomething
# do not match and fall through to the default backend.
#
# Example requests:
#
#   # Create embeddings
#   curl http://localhost:8080/v1/embeddings \
#     -H "Content-Type: application/json" \
#     -d '{"model":"text-embedding-3-small","input":"Hello, world!"}'

listeners:
  - name: ai-gateway
    address: "127.0.0.1:8080"
    filter_chains: [embeddings-routing]

filter_chains:
  - name: embeddings-routing
    filters:
      - filter: router
        routes:
          - path_prefix: "/v1/embeddings"
            cluster: "embeddings-api"
          - path_prefix: "/"
            cluster: "default-backend"

      - filter: load_balancer
        clusters:
          - name: "embeddings-api"
            endpoints:
              - "127.0.0.1:3001"
          - name: "default-backend"
            endpoints:
              - "127.0.0.1:3002"

insecure_options:
  allow_private_endpoints: true # example proxies to local backends