Chain LLM Node

Connect an LLM to a prompt template for text generation.

Written By pvdyck

Last updated 18 days ago

Chain LLM Node

🧪 Labs — Experimental, may change or break without notice


n8n docs	Chain LLM node documentation

What It Does

Sends a prompt to a connected LLM and returns the response. Use it for text generation, summarization, or any task where you need a single LLM call per item.

Connections

Input	Required	Description
Model	Yes	The LLM to call (e.g., OpenRouter, Ollama)
Fallback Model	No	Backup model if the primary fails or hits rate limits
Output Parser	No	Formats the LLM response into structured data

Compatibility

Feature	Status
Sequential item processing	Supported — each item is sent to the LLM one at a time
Batch processing (V1.7+)	Supported — process multiple items in parallel batches for efficiency
Fallback model	Supported — automatically retries with backup model on failure
JSON unwrapping (V1.6+)	Supported — when the LLM returns JSON, it's automatically parsed into fields
Continue on Fail	Supported — failed items don't stop the workflow
Streaming	Not supported — the full LLM response is returned at once
Token counting / cost tracking	Not supported — API costs are tracked at the platform level, not per-node

Error Handling

When the LLM encounters issues, the node provides clear error messages:

Timeout (HTTP 524/504) — "AI model timed out" with retry guidance
Rate limit (HTTP 429) — "AI model rate limited" with wait guidance
Unavailable (HTTP 502/503) — "AI model temporarily unavailable"

If Continue on Fail is enabled, failed items are returned with an error field instead of stopping the workflow.

Status

This integration has been tested with basic operations on indie.money. Some advanced features may not work as expected. Report issues if you encounter them.