n8n

n8n is a workflow automation platform: you wire nodes into flows that move data between services, and AI nodes do the model calls. Its OpenAI nodes accept any OpenAI-compatible endpoint through the credential's Base URL field, so an AI Agent, a chat model, or an embeddings step in n8n can run on Nebul.

Prerequisites

• An n8n instance (n8n Cloud or self-hosted) at a recent version
• A Nebul AI Studio account with API access
• An API key for your active project
• A model ID from the Model Catalog

Configure the credential

n8n's OpenAI credential has a Base URL field that defaults to https://api.openai.com/v1. Point it at Nebul instead:

1. In n8n, open Credentials and create a new OpenAI API credential.
2. Paste your Nebul API key into API Key.
3. Set Base URL to https://api.inference.nebul.io/v1.
4. Save the credential.

Saving runs a connectivity test against GET /v1/models on the base URL, scoped to your key's project. A successful save means the key and endpoint work. The same credential serves the OpenAI node, the OpenAI Chat Model node, and the Embeddings OpenAI node.

Use it in a workflow

Chat model in an AI Agent

1. Add an AI Agent node to your workflow.
2. Attach an OpenAI Chat Model sub-node to it.
3. Select the Nebul credential in the sub-node.
4. Pick a model from the dropdown. n8n loads the list from the endpoint, so you'll only see models enabled for your key's project.

Attach a memory node for multi-turn conversations, then run the workflow. The agent's model calls go to Nebul.

Plain chat completion

For workflows that don't need an agent loop, the OpenAI node with the Message a Model resource sends a chat completion through the same credential.

Embeddings

The Embeddings OpenAI sub-node works the same way. Pick an embedding model from the catalog, for example BAAI/bge-m3, and the vectors come from Nebul.

Troubleshooting

• Credential test fails with 401: the key is wrong, revoked, or from a different project. Create a fresh key in AI Studio and update the credential.
• Credential test fails with 404: the Base URL is missing the /v1 suffix or points somewhere else. It must be https://api.inference.nebul.io/v1 exactly.
• Model dropdown is empty or missing models: /v1/models only returns models enabled for the key's project. Check the project's model access in AI Studio, or type a model ID from the Model Catalog by hand where the node allows it.
• Repeated 401s from a scheduled workflow: a running workflow keeps its saved credential. If you rotated the key, open the credential and update it; the executions after that succeed.

Data and telemetry

n8n collects product telemetry: workflow shapes (not their data), errors, a truncated IP address, and the domains of services you connect to. On n8n Cloud you cannot switch all of it off. On a self-hosted instance, disable diagnostics with an environment variable:

export N8N_DIAGNOSTICS_ENABLED=false

To also stop version-update checks:

export N8N_VERSION_NOTIFICATIONS_ENABLED=false

Workflow data, including prompts you send through nodes, stays on your instance. For how Nebul treats the requests it receives, see Privacy & Security.