Use Autark in the tools you already run.
Autark speaks the OpenAI API, so most tools that work with OpenAI work with Autark. Point the tool at our base URL, swap the key, keep everything else the same. Each guide below covers one tool, including the parts that trip people up.
Need a key first? Sign in to cloud.autark.ai and go to Settings → API Keys.
Pick your tool
n8n
Point the OpenAI Chat Model node at Autark and switch off the Responses API.
Zapier
Use Autark inside any Zap through a webhook step.
Cursor
Override the OpenAI base URL and keep Cursor's chat on Autark.
Hermes
Register Autark as a custom provider in the Hermes config.
OpenCode
Add Autark as an OpenAI-compatible provider in opencode.json.
Claude Code
Needs a translation proxy: Claude Code speaks the Anthropic format.
Connection details
Every integration below uses these three values.
https://api.autark.ai/v1Authorization: Bearer ***GET /v1/modelsThese are the names you pass in the model field:
autark-proautark-agentautark-deepautark-pro-privateautark-agent-privateautark-deep-privateAppend -private to any name for the zero-data-retention variant: same quality, restricted to providers that never retain or train on your prompts. See routing modes for what each one does.
Two formats, and which one you have
Autark serves the OpenAI format. Some tools speak the Anthropic format instead, and those cannot connect directly. The quickest test is to ask the endpoint for a route that does not exist: a 401 means the route exists and only the auth was rejected, while a 404 means the route is not there at all.
OpenAI format
POST /v1/chat/completions. Connects directly. This covers most tools, including n8n, Cursor, Hermes, OpenCode, the OpenAI SDKs, LangChain and Aider.
Anthropic format
POST /v1/messages. Not served. Tools that require it, such as Claude Code, need a translation proxy in between.