Get started

Overview

One OpenAI Responses-compatible API for all models across providers.

Ramp Router puts one OpenAI Responses-compatible endpoint in front of models from multiple providers. Your application holds a single Router key; Router holds the provider credentials and never asks you to send them.

API surface

MethodRoutePurpose
GET/v1/modelsList the models available to the current key
POST/v1/responsesCreate a buffered or streaming model response

Each request sends exactly one of model (a model ID) or models (an ordered fallback list).

Start with the quickstart, or go straight to connecting your app if you already call the Responses API.