AI & LLM Nodes¶

AI nodes integrate with large language models for text generation and analysis.

OpenAI Node¶

Interact with OpenAI's GPT models.

Type¶

n8n-nodes-base.openAi

Parameters¶

Parameter	Type	Required	Default	Description
`apiKey`	string	Yes	-	OpenAI API key
`model`	string	No	`gpt-3.5-turbo`	Model name
`prompt`	string	Yes	-	Input prompt
`maxTokens`	number	No	1000	Max response tokens
`temperature`	number	No	0.7	Randomness (0-2)

Models¶

Model	Description
`gpt-4`	Most capable, slower
`gpt-4-turbo`	Fast GPT-4
`gpt-3.5-turbo`	Fast, cost-effective

Examples¶

Basic Completion¶

{
  "id": "openai-1",
  "name": "Generate Text",
  "type": "n8n-nodes-base.openAi",
  "position": [450, 300],
  "parameters": {
    "apiKey": "={{ $credentials.openai.apiKey }}",
    "model": "gpt-3.5-turbo",
    "prompt": "Summarize this article: {{ $json.content }}",
    "maxTokens": 500
  }
}

Dynamic Prompt¶

{
  "type": "n8n-nodes-base.openAi",
  "parameters": {
    "apiKey": "={{ $credentials.openai.apiKey }}",
    "model": "gpt-4",
    "prompt": "Analyze the following customer feedback and extract:\n1. Sentiment (positive/negative/neutral)\n2. Key topics\n3. Action items\n\nFeedback: {{ $json.feedback }}",
    "maxTokens": 1000,
    "temperature": 0.3
  }
}

Code Generation¶

{
  "type": "n8n-nodes-base.openAi",
  "parameters": {
    "apiKey": "={{ $credentials.openai.apiKey }}",
    "model": "gpt-4",
    "prompt": "Write a {{ $json.language }} function that {{ $json.description }}",
    "maxTokens": 2000,
    "temperature": 0.2
  }
}

Output¶

{
  "json": {
    "response": "Generated text content...",
    "usage": {
      "prompt_tokens": 50,
      "completion_tokens": 150,
      "total_tokens": 200
    },
    "finish_reason": "stop"
  }
}

Field	Description
`response`	Generated text
`usage`	Token usage statistics
`finish_reason`	Why generation stopped

Temperature Guide¶

Temperature	Use Case
0.0 - 0.3	Factual, deterministic
0.4 - 0.7	Balanced creativity
0.8 - 1.2	Creative writing
1.3 - 2.0	Highly random

Anthropic Node¶

Interact with Anthropic's Claude models.

Type¶

n8n-nodes-base.anthropic

Parameters¶

Parameter	Type	Required	Default	Description
`apiKey`	string	Yes	-	Anthropic API key
`model`	string	No	`claude-3-5-sonnet-20241022`	Model name
`prompt`	string	Yes	-	Input prompt
`maxTokens`	number	No	1024	Max response tokens
`temperature`	number	No	1.0	Randomness (0-1)

Models¶

Model	Description
`claude-3-5-sonnet-20241022`	Latest, balanced
`claude-3-opus-20240229`	Most capable
`claude-3-sonnet-20240229`	Fast, capable
`claude-3-haiku-20240307`	Fastest, economical

Examples¶

Basic Completion¶

{
  "id": "anthropic-1",
  "name": "Claude Response",
  "type": "n8n-nodes-base.anthropic",
  "position": [450, 300],
  "parameters": {
    "apiKey": "={{ $credentials.anthropic.apiKey }}",
    "model": "claude-3-5-sonnet-20241022",
    "prompt": "Explain {{ $json.topic }} in simple terms.",
    "maxTokens": 500
  }
}

Analysis Task¶

{
  "type": "n8n-nodes-base.anthropic",
  "parameters": {
    "apiKey": "={{ $credentials.anthropic.apiKey }}",
    "model": "claude-3-opus-20240229",
    "prompt": "Analyze this document and provide:\n1. Summary (2-3 sentences)\n2. Key points\n3. Recommendations\n\nDocument:\n{{ $json.document }}",
    "maxTokens": 2000,
    "temperature": 0.5
  }
}

Data Extraction¶

{
  "type": "n8n-nodes-base.anthropic",
  "parameters": {
    "apiKey": "={{ $credentials.anthropic.apiKey }}",
    "model": "claude-3-sonnet-20240229",
    "prompt": "Extract the following information from this email as JSON:\n- sender_name\n- subject\n- action_items (array)\n- urgency (low/medium/high)\n\nEmail:\n{{ $json.emailBody }}\n\nRespond only with valid JSON.",
    "maxTokens": 500,
    "temperature": 0.1
  }
}

Output¶

{
  "json": {
    "response": "Generated response text...",
    "usage": {
      "input_tokens": 100,
      "output_tokens": 200
    },
    "stop_reason": "end_turn"
  }
}

Common Patterns¶

Prompt Templates¶

Store prompts in variables:

{
  "type": "n8n-nodes-base.set",
  "parameters": {
    "assignments": [
      {
        "name": "prompt",
        "value": "You are a helpful assistant. User query: {{ $json.userInput }}"
      }
    ]
  }
}

Error Handling¶

Check for API errors:

{
  "type": "n8n-nodes-base.filter",
  "parameters": {
    "conditions": [
      {
        "leftValue": "={{ $json.error }}",
        "operator": "notExists"
      }
    ]
  }
}

Response Parsing¶

Parse JSON from AI response:

{
  "type": "n8n-nodes-base.code",
  "parameters": {
    "language": "javascript",
    "code": "const response = items[0].json.response;\ntry {\n  return [{json: JSON.parse(response)}];\n} catch(e) {\n  return [{json: {raw: response, parseError: true}}];\n}"
  }
}

Quick Reference¶

Node	Type	Models
OpenAI	`n8n-nodes-base.openAi`	GPT-4, GPT-3.5-turbo
Anthropic	`n8n-nodes-base.anthropic`	Claude 3.5, Claude 3

Use Cases¶

Use Case	Recommended	Model
Summarization	Either	Sonnet/GPT-4
Code generation	OpenAI	GPT-4
Analysis	Anthropic	Opus
Quick responses	Either	Haiku/GPT-3.5
Data extraction	Either	Sonnet/GPT-4