Integrations
Point existing OpenAI, Anthropic, and Gemini tools at orientx
Most integrations are the same swap: set the base URL and the API key, then pick a model id from
GET /v1/models. Some clients speak Anthropic Messages or the Gemini API instead of Chat
Completions — same key, different host path. Each page spells out which one to use, and links to
that client's official docs for the same settings.
| Protocol | Base URL | Auth |
|---|---|---|
OpenAI (/v1/chat/completions, /v1/responses) | https://api.orientx.ai/v1 | Authorization: Bearer <ORIENTX_API_KEY> |
Anthropic (/v1/messages) | https://api.orientx.ai | Bearer or x-api-key |
Gemini (/v1beta/models/{model}:{action}) | https://api.orientx.ai | x-goog-api-key or ?key= |
Create a key in the console first. Support for tools, vision, and
the Responses API varies by model — if a client asks for a feature the model cannot honor, the API
returns 400 rather than ignoring the field.
Libraries
OpenAI SDK
Official Python and TypeScript clients. The shortest path if you already call OpenAI.
CrewAI
Multi-agent crews via an OpenAI-compatible LLM object.
Coding tools
Claude Code
Anthropic's CLI. Set ANTHROPIC_BASE_URL to the Messages API host.
Codex CLI
OpenAI's coding agent. Add orientx as a custom model provider.
Gemini CLI
Google's CLI. Set GOOGLE_GEMINI_BASE_URL — do not patch install files.
OpenCode
Terminal agent. Anthropic, OpenAI, or Gemini as a custom provider.
Hermes Agent
Nous Research's CLI on Chat Completions (api_mode required).
Pi
Earendil's agent. Use anthropic-messages so tool calls work.
Cursor
Override the OpenAI base URL in Cursor's model settings.
Continue
VS Code / JetBrains autocomplete and chat.
Aider
Terminal pair-programming against an OpenAI-compatible endpoint.