Azure OpenAI Response

AI Assistants automation

v1.0.4

Last updated Oct 10, 2025

n8n community node for Azure OpenAI Response API

4 Weekly Downloads

46 Monthly Downloads

View on NPM GitHub Repository

Included Nodes

Azure OpenAI Response

Description

n8n-nodes-azure-openai-responses

This is an n8n community node that integrates the Azure OpenAI Response API into your n8n workflows.

The Response API is a new stateful API from Azure OpenAI that combines the best capabilities from chat completions and assistants APIs in one unified experience. It supports advanced features like stateful conversations, tool calling, reasoning models (o1, o3, o4 series), background processing, and more.

n8n is a fair-code licensed workflow automation platform.

Disclaimer

This is an unofficial community node and is not affiliated with, endorsed by, or supported by Microsoft Corporation or n8n GmbH.

Azure, Azure OpenAI, and related trademarks are property of Microsoft Corporation. Users must comply with Microsoft's Azure OpenAI Service terms and conditions.

This package is provided "as is" under the MIT License without warranty of any kind.

Installation
Features
Credentials
Usage
Parameters
Multiple Outputs
Examples
Resources
Version History

Installation

Follow the installation guide in the n8n community nodes documentation.

npm

npm install n8n-nodes-azure-openai-responses

Manual Installation (Development)

# Clone this repository
git clone https://github.com/mlangcode/n8n-nodes-azure-openai-responses.git
cd n8n-nodes-azure-openai-responses

# Install dependencies and build
npm install
npm run build

# Link to your local n8n
npm link
cd ~/.n8n
npm link n8n-nodes-azure-openai-responses

# Restart n8n

Features

✅ Simple & Focused: Single operation (Create Response) – no unnecessary complexity
✅ Multiple Outputs: Separate outputs for messages, tool calls, and full responses
✅ Agent-Ready: Route tool calls to functions and return results seamlessly
✅ Flexible Input: Accept strings, message arrays, or webhook data
✅ Dynamic Tools: Pass tool definitions from webhooks or external sources
✅ Stateful Conversations: Multi-turn conversations with automatic context
✅ Built-in Tools: Code interpreter, file search, image generation
✅ Reasoning Models: Support for o1, o3, o4 series with configurable effort
✅ Error Handling: Comprehensive error messages and validation
✅ Long-Running Requests: Handles reasoning models (n8n supports up to 1 hour timeout)

Credentials

This node uses Azure OpenAI credentials with the following fields:

Resource Name: Your Azure OpenAI resource name (e.g., my-resource-name)
API Key: Your Azure OpenAI API key
API Version: The API version to use (default: 2025-04-01-preview)

The base URL is automatically constructed as: https://{resourceName}.openai.azure.com/openai/v1

Setting Up Credentials

In n8n, go to Credentials → New
Search for "Azure OpenAI API"
Fill in your resource name and API key
Click Save

Usage

Basic Usage

Add the "Azure OpenAI Response" node to your workflow
Configure your Azure OpenAI credentials
Set the model deployment name (e.g., gpt-4.1-mini)
Enter your input/prompt
Configure additional parameters as needed

The node subtitle will display: Model (Model Name) on Resource (Resource Name) for easy identification.

Stateful Conversations

To continue a multi-turn conversation:

Store the responseId from the previous response
In the next request, set Previous Response ID in Additional Fields
The API will maintain conversation context automatically

Parameters

Required Parameters

Model: The deployment name of your model (e.g., gpt-4.1-mini, gpt-4o, o3-mini)
Input: The input for the model
- String: "Tell me a joke"
- Expression: {{ $json.messages }}
- Message array: [{"role": "user", "content": "Hello"}]

Tools (Dynamic)

Pass tool definitions from webhooks or external sources!

Format Options:

Full OpenAI format:

[
  {
    "type": "function",
    "function": {
      "name": "get_weather",
      "description": "Get current weather",
      "parameters": {
        "type": "object",
        "properties": {
          "city": {"type": "string"}
        },
        "required": ["city"]
      }
    }
  }
]

Just function definitions (auto-wrapped):

[
  {
    "name": "get_weather",
    "description": "Get current weather",
    "parameters": {...}
  }
]

Using expressions:

{{ $json.tools }}

Use Case: When building conversational AI agents that receive tool definitions dynamically from webhooks or API calls.

Additional Fields

Conversation & State

Instructions: System instructions for the assistant (similar to system messages)
Previous Response ID: Continue from a previous response for stateful conversations
Store Conversation: Whether to store conversation history (default: true)
Metadata: Custom JSON metadata for tracking

Output Control

Temperature (0-2): Controls randomness in outputs (default: 1.0)
Top P (0-1): Nucleus sampling parameter (default: 1.0)
Max Output Tokens: Maximum tokens in response (0 for model default)

Reasoning Models (o1, o3, o4 series)

Reasoning Effort: Choose between low, medium, or high (default: medium)
- Higher effort provides more thorough reasoning but takes longer
- Reasoning models may take several minutes to respond

Advanced Options

Background Mode: Run the request asynchronously (requires polling – advanced use)
Stream: Enable streaming responses (not recommended for standard workflows)
Include Fields: Additional fields to include (e.g., reasoning.encrypted_content)

Tools Configuration

Built-in Tools

Select from the dropdown:

Code Interpreter: Execute Python code in a sandboxed environment
File Search: Search through uploaded files
Image Generation: Generate images using DALL-E

Custom Functions

Define custom functions using JSON format:

{
  "name": "get_weather",
  "description": "Get the current weather for a location",
  "parameters": {
    "type": "object",
    "properties": {
      "location": {
        "type": "string",
        "description": "City name"
      }
    },
    "required": ["location"]
  }
}

Tool Options

Tool Choice: Control how tools are used
- auto: Let the model decide (default)
- required: Force the model to use a tool
- none: Don't use any tools
Parallel Tool Calls: Allow multiple tools to be called simultaneously (default: true)

Multiple Outputs

The node has three outputs for flexible workflow routing:

Output 0: Messages 💬

When: Model returns a text/message response
Contains: Formatted message data

{
  "responseId": "resp_123",
  "outputText": "The answer is 42",
  "messages": [
    {
      "id": "msg_123",
      "role": "assistant",
      "content": [{"type": "output_text", "text": "The answer is 42"}]
    }
  ],
  "usage": {...}
}

Output 1: Tools 🔧

When: Model requests tool/function calls
Contains: Tool call definitions

{
  "responseId": "resp_123",
  "toolCalls": [
    {
      "id": "call_123",
      "type": "function",
      "function": {
        "name": "get_weather",
        "arguments": "{\"city\":\"London\"}"
      }
    }
  ]
}

Use this to:

Route to function execution nodes
Execute the requested tools
Return results back to the AI

Output 2: Full Response 📄

Always active – Contains the complete API response

{
  "responseId": "resp_123",
  "status": "completed",
  "outputText": "...",
  "messages": [...],
  "toolCalls": [...],
  "fullResponse": {
    "id": "resp_123",
    "model": "gpt-4.1-mini",
    "output": [...],
    "usage": {...},
    // ... complete API response
  },
  "usage": {...}
}

Use this when:

You need to pass the entire response to another application
Debugging/logging full API responses
Accessing raw response fields not exposed in other outputs

Error Handling

When errors occur (and "Continue on Fail" is enabled):

Error details are sent to Output 2 (Full Response)
Includes HTTP status codes and error messages
Workflow continues instead of stopping

Examples

Example 1: Simple Text Generation

Model: gpt-4.1-mini
Input: "Explain quantum computing in simple terms"

Output: Goes to Output 0 (Messages)

Example 2: Stateful Conversation

First Request:

Model: gpt-4.1-mini
Input: "What is the capital of France?"

Second Request (using responseId from first):

Model: gpt-4.1-mini
Input: "What is its population?"
Additional Fields > Previous Response ID: {{ $json.responseId }}

Example 3: Agent with Tool Calling

Workflow:

[Webhook Trigger]
  Receives: {
    "messages": [{"role": "user", "content": "What's the weather in Paris?"}],
    "tools": [{"name": "get_weather", "parameters": {...}}]
  }
    ↓
[Azure OpenAI Response Node]
  Input: {{ $json.messages }}
  Tools (Dynamic): {{ $json.tools }}
    ↓
  Output 1 (Tools) ──→ [HTTP Request: Call Weather API]
    ↓
  Return result to AI with responseId

Example 4: Reasoning Model

Model: o3-mini
Input: "Solve this logic puzzle: If all A are B, and some B are C, what can we conclude about A and C?"
Additional Fields > Reasoning Effort: high

Note: May take 1-3 minutes for complex reasoning.

Example 5: Dynamic Tools from Webhook

Webhook Body:

{
  "user_message": "Book a flight to Tokyo",
  "available_functions": [
    {
      "type": "function",
      "function": {
        "name": "search_flights",
        "description": "Search for flights",
        "parameters": {
          "type": "object",
          "properties": {
            "destination": {"type": "string"},
            "date": {"type": "string"}
          }
        }
      }
    }
  ]
}

Node Configuration:

Input: {{ $json.user_message }}
Tools (Dynamic): {{ $json.available_functions }}

Example 6: Code Interpreter

Model: gpt-4o
Input: "Create a bar chart showing sales data: Jan 100, Feb 150, Mar 200"
Tools > Built-in Tools > Tool Type: Code Interpreter

Output: Code execution results in Output 1 (Tools)

Resources

Compatibility

Requires n8n version 1.60.0 or later
Compatible with all Azure OpenAI Response API supported models
Supported regions: australiaeast, eastus, eastus2, francecentral, japaneast, norwayeast, polandcentral, southindia, swedencentral, switzerlandnorth, uaenorth, uksouth, westus, westus3

Supported Models

The Response API supports the following model families:

GPT-4.1 Series: gpt-4.1, gpt-4.1-mini, gpt-4.1-nano
GPT-4o Series: gpt-4o, gpt-4o-mini
GPT-5 Series: gpt-5, gpt-5-mini, gpt-5-nano, gpt-5-chat, gpt-5-codex
o-Series (Reasoning): o1, o3, o3-mini, o4-mini
Image Generation: gpt-image-1
Computer Use: computer-use-preview

Note: Not all models are available in all regions. Check the Azure OpenAI models page for regional availability.

Limitations

Web search tool is not currently supported by Azure OpenAI Response API
Image generation with multi-turn editing and streaming is coming soon
Images cannot be uploaded as files and referenced as input (coming soon)
PDF input is supported, but setting file upload purpose to user_data is not yet supported
Background mode may have higher latency initially

Troubleshooting

"Model not found" error

Verify your deployment name matches exactly
Ensure the model is deployed in your Azure OpenAI resource

"Invalid API key" error

Check that your API key is correct
Verify the resource name is correct

"Input is required" error

Ensure the Input field is not empty
If using expressions, verify they resolve to a value

"Unexpected end of JSON input"

Use string type fields (Input, Tools Dynamic) with expressions like {{ $json.messages }}
Don't leave JSON type fields with invalid JSON

Reasoning models taking too long

Reasoning models (o1, o3, o4) can take 1-5 minutes
n8n workflow timeout default is 1 hour (sufficient for most cases)
Consider setting reasoning effort to "low" or "medium" for faster responses

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT

Version History

1.0.0

Initial release with streamlined single operation (Create Response)
Three outputs: Messages, Tool Calls, and Full Response
Support for dynamic tool definitions from webhooks
Flexible input handling (strings, message arrays, expressions)
Comprehensive error handling with detailed messages
Full parameter support including tools, reasoning, and background processing
Support for stateful conversations
Built-in tools: code interpreter, file search, image generation
Subtitle display showing model and resource name

Author

mlangcode

Support

For issues, questions, or contributions, please visit the GitHub repository.