OpenAI connection

An account with OpenAI, or with any endpoint that speaks the same protocol, for AI steps to run on. In the connection picker it is OpenAI.

Powers these nodes

No OpenAI-specific nodes yet. It serves the built-in AI Agent action, the AI Judgment condition, and your default model—anywhere a Provider Account is chosen.

Fields

Field	Required	Description
Name	Yes	A name for the account. It labels the account everywhere it is offered, so say whose account it is.
Base URL	No	The endpoint's OpenAI-compatible API root—a vendor's own, an aggregator's, or a gateway you host. Leave it blank for OpenAI's own, which the placeholder shows.
API Key	Yes	The key this endpoint authenticates with. Stored encrypted and never logged.

Because the base URL is a field, one connection type covers four situations: OpenAI itself, a vendor's compatible API, an aggregator whose single key reaches many vendors, and an endpoint you run yourself.

Compatibility

Some endpoints speak the protocol but not all of it. The Compatibility tab carries two switches for that:

Switch	Turn it on when
Don't Use response_format	The endpoint cannot return an answer in a shape you asked for. Steps that need structured output ask the model to call a function instead, which works anywhere function calling does. The symptom of getting this wrong is a step failing because the model returned no structured output, or the wrong shape.
Don't Use response_format Together With Tools	The endpoint handles a shaped answer on its own but rejects a request that asks for one and offers tools. Only those requests switch approach. The symptom of getting this wrong is an error on tool-using steps alone.

Leave both off for OpenAI itself.

Connecting

Credential-based: there is no Connect button. Paste the key and save.

The Model field wherever you choose one then offers what this endpoint lists. Not every compatible endpoint publishes a catalog, so you can also type an identifier the list doesn't show.