Integration overview
Connect the OpenAI SDK and compatible clients.
Prepare connection details
Prepare the gateway URL, XHub virtual key and public model name. Use the same gateway root Base URL as the playground; do not append /v1. Field names depend on tool version.
Choose an integration
Python applications can use the SDK example. Tools supporting custom OpenAI endpoints can use client configuration. Tools restricted to fixed provider URLs cannot directly use a custom gateway.
Verify compatibility
List models and send a minimal chat request. Responses, images, audio and streaming require matching deployment protocols. Connection success does not prove every feature is supported.
Choose the protocol
| Tool | Protocol | Base URL |
|---|---|---|
| OpenAI SDK, Cursor, Cline, Chatbox | Chat completions | Playground gateway root |
| Codex CLI | Responses | Playground gateway root |
| Claude Code | Messages | Root without /v1 |
| OpenCode and agent frameworks | Selected adapter | Check provider settings |
Each tool has a dedicated guide in the directory. XHub has no third-party automatic setup installer. A successful model list does not verify every model capability.