Description
@sevenlab/n8n-nodes-litellm-chat-model
An n8n community node for LiteLLM — a proxy that provides a unified OpenAI-compatible API across 100+ LLM providers (OpenAI, Anthropic, Gemini, Azure, Ollama, and more).
Instead of managing separate credentials and node configurations per provider, point this node at your LiteLLM proxy and pass model parameters as plain JSON.
n8n is a fair-code licensed workflow automation platform.
Screenshots
Workflow: LiteLLM Chat Model as AI Agent sub-node
!n8n workflow showing LiteLLM Chat Model connected to an AI Agent node
Node configuration
Installation
Follow the installation guide in the n8n community nodes documentation.
Or install directly:
@sevenlab/n8n-nodes-litellm-chat-model
Credentials
You need a running LiteLLM proxy. Configure the following:
- Base URL — the base URL of your LiteLLM proxy (e.g.
https://your-litellm-proxy.example.com/v1) - API Key — your LiteLLM API key
Usage
1. Add the LiteLLM Chat Model sub-node to an AI Agent or Chain node
2. Select your LiteLLM account credential
3. Pick a Model from the dropdown — loaded dynamically from your proxy’s /models endpoint
4. Optionally configure parameters under Options
5. Use Custom Parameters (JSON) for any provider-specific fields not covered by the options above
Options
The following parameters are available as structured fields:
| Option | Description |
|—|—|
| Frequency Penalty | Penalise repeated tokens (-2 to 2) |
| Max Retries | Number of retries on API errors (default: 2) |
| Maximum Completion Tokens | Upper bound including reasoning tokens |
| Maximum Number of Tokens | Max tokens for the completion |
| Presence Penalty | Penalise tokens already present (-2 to 2) |
| Reasoning Effort | Controls thinking token budget: low, medium, high, xhigh, max, none, disable |
| Response Format | text or json_object |
| Sampling Temperature | 0–2, higher = more random (default: 0.7) |
| Seed | For deterministic outputs |
| Stop Sequences | JSON array of stop strings, e.g. ["n", "END"] |
| Timeout | Request timeout in ms (default: 60000) |
| Top K | Gemini/Vertex only |
| Top P | Nucleus sampling probability (0–1) |
| User ID | End-user identifier (OpenAI / Anthropic) |
Custom Parameters (JSON)
Passes any extra fields directly to the API, overriding the structured options above. Use this for provider-specific parameters not covered by the options, for example:
{ "safety_settings": [...] }
{ "citations": true }
For the full list of supported parameters per provider, see:
Compatibility
Minimum n8n version: 2.10.0
