Skip to main content

OpenAI

Where to issue an OpenAI API key and connect it to Nora.
OpenAI integrations let your agents call GPT models, generate embeddings, render images via DALL-E, and transcribe audio with Whisper.

Where to apply for credentials

1

Sign in

2

Create a project key

Settings → API Keys → Create new secret key. Pick Project scoping for production agents (revocable per project) or Service account for shared infra agents.
3

Copy the key

The key starts with sk-.... OpenAI shows it only once.

(Optional) Org / project IDs

If you’re on a multi-org account, find the Organization ID under Settings → Organization → General. Project IDs live under Project Settings. Setting these on the integration is mostly useful for cost attribution.

Connect in Nora

Paste the API key into the API Key field, optionally set Organization ID and a default model (e.g. gpt-4o-mini), and click Connect. Nora calls GET https://api.openai.com/v1/models and reports how many models are visible to the key.

Verify the connection

The Test button calls /v1/models. 401 means the key was revoked; 429 means you’re rate-limited (try again).

MCP server

No official OpenAI MCP server today. The standard env var (OPENAI_API_KEY) is exposed to the agent so any community OpenAI MCP can pick it up.

Environment variables Nora injects

VariableSource
OPENAI_API_KEYAPI Key field
OPENAI_ORG_IDOrganization ID (if set)
OPENAI_MODELDefault Model (if set)