Package Information
Documentation
n8n-nodes-oci-generative-ai
An n8n community node that connects Oracle Cloud Infrastructure (OCI) Generative AI as a native LLM chat model — compatible with the AI Agent, LangChain chains, and any node that accepts a Language Model sub-node.
What it does
This node exposes OCI Generative AI's OpenAI-compatible inference endpoint (/v1/chat/completions) as a proper n8n LangChain chat model. You can plug it directly into the AI Agent node (Tools Agent) or any chain node, the same way you would use OpenAI, Anthropic, or Google Gemini nodes.
Key capabilities:
- Tool calling (function calling) — the AI Agent can use tools (HTTP requests, database queries, custom code, etc.) with models that support it
- Streaming — yields response chunks progressively when used with the Chat Trigger
- JSON mode — optionally forces the model to return a valid JSON object
- Full LangChain integration — works with memory nodes (Postgres Chat Memory, Redis, etc.), output parsers, and chains
Why use this instead of a generic HTTP node?
| Feature | Generic HTTP Node | This node |
|---|---|---|
| Works as AI Agent LLM | No | Yes |
| Tool calling support | No | Yes |
| LangChain memory integration | No | Yes |
| Streaming output | No | Yes |
| Credential manager | Manual headers | Built-in |
| OpenAI-compatible abstraction | Manual | Automatic |
Supported regions & models
OCI Generative AI is available in multiple regions. Use the Base URL that matches your OCI tenancy:
| Region | Base URL |
|---|---|
| Brazil East (São Paulo) | https://inference.generativeai.sa-saopaulo-1.oci.oraclecloud.com/20231130/actions |
| US Midwest (Chicago) | https://inference.generativeai.us-chicago-1.oci.oraclecloud.com/20231130/actions |
| Germany Central (Frankfurt) | https://inference.generativeai.eu-frankfurt-1.oci.oraclecloud.com/20231130/actions |
| UK South (London) | https://inference.generativeai.uk-london-1.oci.oraclecloud.com/20231130/actions |
Model availability varies by region. Examples:
- São Paulo:
meta.llama-3.1-8b-instruct,meta.llama-3.1-70b-instruct,cohere.command-r-plus - Chicago:
meta.llama-3.3-70b-instruct,google.gemini-2.5-flash,cohere.command-a-03-2025
Check the OCI Generative AI documentation for the latest model list per region.
Installation
Via n8n Community Nodes (recommended)
- In your n8n instance, go to Settings → Community Nodes
- Click Install
- Enter
n8n-nodes-oci-generative-ai - Click Install
Requires n8n v1.0.0 or later with community nodes enabled.
Via npm (self-hosted / Docker)
npm install n8n-nodes-oci-generative-ai
Then set the environment variable so n8n loads the package:
N8N_CUSTOM_EXTENSIONS=n8n-nodes-oci-generative-ai
Configuration
Credentials
Create a new Oracle OCI Generative AI Inference API credential with:
| Field | Description |
|---|---|
| Base URL | The OCI inference endpoint for your region (see table above) |
| Bearer Token | Your OCI authentication token (from oci session authenticate or a service account) |
| Test Model | Model used only for the credential connection test (default: meta.llama-3.1-8b-instruct) |
Node parameters
| Parameter | Description |
|---|---|
| Model | The model ID to use (e.g. meta.llama-3.1-70b-instruct) |
| Temperature | Controls randomness (0 = deterministic, 1 = creative) |
| Max Tokens | Maximum number of tokens in the response |
| Top P | Nucleus sampling parameter |
| Response Format | Default (text) or JSON Object (forces JSON output) |
Usage example
- Add an AI Agent node to your workflow
- Connect OCI Generative AI Chat Model as the Language Model sub-node
- (Optional) Connect a Memory node (e.g. Postgres Chat Memory) to maintain conversation history
- (Optional) Add Tool sub-nodes (HTTP Request, Code, etc.) for agentic behavior
The agent will automatically use tool calling to invoke the connected tools when appropriate.
Requirements
- n8n >= 1.0.0
- An active Oracle Cloud Infrastructure account with Generative AI access
- A valid OCI Bearer Token for the inference API