Introduction

Connect to different AI models through one API gateway.

API is an AI API gateway with a unified model endpoint, API key management, channel configuration, and request logs.

This documentation is for developers using an existing deployment. Your service address, available models, and account permissions depend on that deployment.

Start with your first request

Follow the quickstart to prepare your service address and API key, list available models, and send a chat request with cURL, Python, or TypeScript.

Understand the three addresses

Name	Example	Purpose
Console URL	https://gateway.example.com	Sign in, create API keys, and inspect request logs
API base URL	https://gateway.example.com/v1	Configure a client or SDK
Chat endpoint	https://gateway.example.com/v1/chat/completions	Send a Chat Completions request

These are example addresses. Use the actual API address provided by your administrator; the console and API may use different domains or path prefixes. The documentation website is not your API endpoint.

Models and protocols

Supported protocols and capabilities vary by model. The quickstart uses the OpenAI-compatible Chat Completions endpoint, so select a chat model that supports this endpoint and is accessible with your API key.

Image, audio, and other specialized models in the model list may not support chat requests. Refer to your deployment administrator and the model description for supported capabilities.

切换到简体中文