R9S is a distributed AI gateway. Applications send model requests to R9S, and R9S forwards those requests through the best available provider path according to compatibility, model availability, regional health, fallback rules, and account policy.
Gateway
The gateway is the runtime boundary for AI traffic. It receives requests at https://gw.r9s.ai/v1, authenticates the caller, normalizes OpenAI-compatible or Anthropic-compatible payloads, selects an upstream route, and records usage and telemetry.
Use the gateway when you want application code to depend on one endpoint instead of many provider-specific integrations.
Model identifiers
Each request names the model it wants to use:
{
"model": "gpt-4o-mini"
}Model identifiers are account-visible names such as gpt-4o-mini, qwen-plus, claude-sonnet-4.5, or embedding and audio model names. Call GET /models to confirm the exact set available to your account.