Google Cloud API Gateway adds model routing in public preview

8/4/2026

Google Cloud API Gateway model routing sends AI requests to Gemini, Claude, and more Avoid hardcoded endpoints and proxies while adding rate limits, token tracking, and control

Google Cloud has introduced model routing in public preview for API Gateway, giving developers a way to send AI requests to different models without hardcoding endpoints or running separate proxies. The update lets the gateway accept OpenAIcompatible requests and route them dynamically to backends such as Gemini, Claude, or OpenAI OSSGPT. Google says the feature can be used on its own for tasks like rate limiting and token tracking, or alongside the Gemini Enterprise Agent Platform for more controlled traffic handling. To set it up, developers define routing rules in an OpenAPI 3.x specification, deploy the updated gateway configuration, and then send standard requests to the gateway endpoint. The gateway transcodes the request to the backend’s native format and forwards it based on the configured model router. Google says all backends within a single router must share the same host, and the feature is available now in public preview.