API Basics

All inference APIs use the gateway Base URL configured for this deployment. Current Base URL: http://127.0.0.1:11113

Base URL and API key

Create a key in Token Management and keep it in server-side secret storage. The same platform key can be used with every client protocol currently supported by the gateway; the header or SDK option used to carry it depends on the selected protocol.

The code examples display this address directly. See Protocols and SDK Compatibility for protocol-specific authentication, endpoint, stream, and error formats.

Choose an endpoint

Choose the endpoint from three pieces of information:

1. The task you want to perform, such as text generation, embedding, image processing, or video processing.
2. The client protocol used by your SDK or existing application.
3. The operations and execution modes supported by the selected model.

Model IDs and managed routing

Use the complete public model ID shown on the Models page. A public model ID identifies the model exposed by the platform; it does not reveal or select an internal upstream channel.

Provider channels, health checks, weighting, and failover are managed by the platform. Top-level provider, routing, and route fields are rejected.

Request tracing

Business responses include x-request-id. Record it with your application logs and include it when investigating a failure or contacting support. For response timing and task behavior, continue with Synchronous, Streaming, and Asynchronous Requests.