Integrations

Router9 is fully compatible with the OpenAI API format. Any AI harness that speaks that format works with Router9 out of the box — usually with nothing more than a base URL and an API key.

Quick Setup

Most harnesses only need a base URL and API key:

OPENAI_BASE_URL=https://api.router9.com/v1
OPENAI_API_KEY=sk-r9k-your-key-here

Router9 model ids are provider-prefixed (for example anthropic/claude-opus-5 or deepseek/deepseek-v4-pro), not short aliases like gpt-4o. Always pull the exact ids your deployment serves from GET /v1/models — requesting an id that isn't configured returns 404 model_not_found.

Supported harnesses

Each harness has a dedicated guide with its exact configuration:

Harness	Setup
Hermes	Custom endpoint in ~/.hermes/config.yaml
OpenClaw	Custom provider in openclaw.json
Goose	Custom OpenAI provider in ~/.config/goose/custom_providers/
opencode	Custom provider in opencode.json
Letta	OPENAI_API_BASE on the Letta server
Nanobot	Custom provider in ~/.nanobot/config.json
Claude Code	MCP tools only — model routing pending
Codex	MCP tools only — model routing pending

Using a harness that isn't listed? If it supports OpenAI-compatible endpoints, the Quick Setup above is all you need — set the base URL and your sk-r9k- key.