Generate Text

Generates text completions using OpenAI's language models with support for vision, chat history, tool calling, and streaming responses.

Common Properties

• Name - The custom name of the node.
• Color - The custom color of the node.
• Delay Before (sec) - Waits in seconds before executing the node.
• Delay After (sec) - Waits in seconds after executing node.
• Continue On Error - Automation will continue regardless of any error. The default value is false.

info

If the ContinueOnError property is true, no error is caught when the project is executed, even if a Catch node is used.

Inputs

• Connection Id - Connection identifier from Connect node. Optional if providing API Key directly.
• API Key - OpenAI API key credential. Optional if using Connection Id.
• System Prompt - Instructions that guide the AI's behavior and personality. Default: "You are a helpful assistant."
• User Prompt - The user's message or question to send to the AI model. Required.
• Chat History - Array of previous conversation messages for maintaining context across multiple interactions.
• Tool Message - Tool response message to include in the conversation (for function calling workflows).
• Image Url - URL or local file path to a single image for vision capabilities.
• Images - Array of image URLs or local file paths for multi-image analysis.

Options

Model Selection

• Model - Language model to use:
  • GPT-5 - Latest and most capable model
  • GPT-5 Mini - Fast and cost-effective GPT-5 variant
  • GPT-5 Nano - Ultra-lightweight GPT-5 variant
  • GPT-5 Pro - Enhanced GPT-5 for complex tasks
  • GPT-5 Chat - Optimized for conversational tasks
  • o3 - Advanced reasoning model
  • o3 Mini - Compact reasoning model
  • o4 Mini - Next-gen compact reasoning model
  • GPT-4.1 - Latest GPT-4 generation
  • GPT-4.1 Mini - Efficient GPT-4.1 variant
  • GPT-4.1 Nano - Ultra-compact GPT-4.1
  • GPT-4o - Multimodal model with vision
  • GPT-4o Mini - Efficient multimodal model
  • GPT-4o Audio Preview - Model with audio capabilities
  • o1 - Reasoning model
  • o1 Mini - Compact reasoning model
  • o1 Pro - Professional reasoning model
  • GPT-3.5 Turbo - Fast and affordable
  • Custom Model - Specify a custom model name
• Custom Model - Custom model name when "Custom Model" is selected.
• Use Robomotion AI Credits - Use Robomotion credits instead of your own API key.

Generation Settings

• Number of Generations - Number of text responses to generate (1-4). Default: 1.
• Stream - Enable streaming response for real-time token generation. Default: false.
• JSON Mode - Force the model to output valid JSON. Default: false.
• Temperature - Controls randomness (0.0-2.0). Lower values are more focused and deterministic. Higher values are more creative.
• Top P - Alternative to temperature using nucleus sampling (0.0-1.0). Don't use both temperature and top_p.
• Max Tokens - Maximum number of tokens to generate in the response.
• Frequency Penalty - Penalizes tokens based on their frequency in the text so far (-2.0 to 2.0). Reduces repetition.
• Presence Penalty - Penalizes tokens based on whether they appear in the text so far (-2.0 to 2.0). Encourages topic diversity.
• Stop Sequences - Up to 4 sequences where the model will stop generating tokens.

Reasoning (o-series models)

• Reasoning Effort - Computational effort for reasoning models (o1, o3, o4-mini):
  • Low - Faster, less thorough
  • Medium - Balanced (default)
  • High - Slower, more thorough

Vision Settings

• Image Detail - Image analysis detail level for vision models:
  • Auto - Model decides based on image (default)
  • High - Detailed analysis, higher cost
  • Low - Basic analysis, lower cost

Tool Calling

• Tools Function - Function definition object with name, description, and parameters schema for tool calling.
• Tool Choice - Control which tool the model calls (auto, none, or specific function).

Advanced

• Seed - Random seed for reproducible outputs (Beta feature).
• User - Unique identifier for your end-user (for monitoring and abuse prevention).
• Timeout (seconds) - Request timeout in seconds. Default: 120.
• Include Raw Response - Include the full API response object in outputs. Default: false.

Outputs

• Text - Generated text response. Returns a string if Number of Generations is 1, or an array of strings if more than 1.
• Raw Response - Complete API response object (only set when "Include Raw Response" is enabled).

How It Works

The Generate Text node sends your prompt to OpenAI's language models. When executed:

1. Validates the connection or API key
2. Builds the conversation with system prompt, user prompt, and optional chat history
3. Processes any images (encodes local files to base64 or uses URLs directly)
4. Configures all generation parameters
5. Sends the request to the selected model
6. Returns the generated text response

Usage Examples

Example 1: Simple Text Generation

Input:
- User Prompt: "Write a professional email asking for a meeting next week"
- Model: gpt-4o-mini

Output:
- Text: "Subject: Meeting Request for Next Week..."

Example 2: Vision - Analyze an Image

Input:
- User Prompt: "What objects are in this image?"
- Image Url: "C:/screenshots/photo.jpg"
- Model: gpt-4o

Output:
- Text: "I can see a laptop, coffee mug, notebook, and pen on a wooden desk..."

Example 3: JSON Mode for Structured Output

Input:
- System Prompt: "Extract contact information as JSON"
- User Prompt: "John Doe, email: john@example.com, phone: 555-1234"
- JSON Mode: enabled
- Model: gpt-4o

Output:
- Text: {"name": "John Doe", "email": "john@example.com", "phone": "555-1234"}

Example 4: Multi-turn Conversation

Flow:
1. First call:
   - User Prompt: "What is the capital of France?"
   - Save response to msg.history

2. Second call:
   - User Prompt: "What is its population?"
   - Chat History: msg.history
   - Model understands "its" refers to Paris

Example 5: Temperature Control

Creative writing (high temperature):
- Temperature: 1.5
- User Prompt: "Write a creative story opening"
- Result: More varied, creative responses

Factual extraction (low temperature):
- Temperature: 0.2
- User Prompt: "Extract the date from: Meeting on Jan 15"
- Result: More consistent, focused responses

Requirements

• Either a Connection Id from Connect node OR an API Key credential
• Non-empty User Prompt
• For vision: GPT-4o or GPT-4o-mini model
• For JSON mode: Compatible model (GPT-4o, GPT-4-turbo, etc.)
• For tool calling: Model with function calling support
• Local images must be in JPEG, PNG, WebP, or GIF format (max 20MB)

Error Handling

The node will return errors in these cases:

• ErrInvalidArg: User Prompt is empty, invalid model selection, or custom model name not provided
• ErrAPICall: OpenAI API errors (rate limits, invalid requests, model not available)
• File errors: Image file not found or unsupported format
• Timeout errors: Request exceeded timeout duration

Tips for RPA Developers

• Model Selection: Use gpt-4o-mini for most tasks - it's fast and affordable. Use GPT-5 or o3 for complex reasoning.
• Temperature: Use 0-0.3 for factual tasks, 0.7-1.0 for creative tasks, 1.5+ for very creative outputs.
• Token Management: Set Max Tokens to prevent excessive costs. Monitor usage via the raw response.
• Vision: Use "low" detail for simple image analysis to save costs. Use "high" for detailed analysis.
• Chat History: Build conversations by passing previous responses as chat history. Format: [{role: "user", content: "..."}, {role: "assistant", content: "..."}]
• JSON Mode: Always instruct the model to output JSON in your prompt when using JSON mode.
• Streaming: Enable for real-time UIs. Disable for batch processing.
• Error Handling: Wrap in Try-Catch blocks and handle rate limit errors with retries.
• Cost Optimization: Use lower-cost models for simple tasks, cache system prompts, and use max_tokens to limit costs.

Common Errors and Solutions

Error: "User Prompt cannot be empty"

• Solution: Ensure the User Prompt input has a value. It cannot be blank.

Error: "Custom Model cannot be empty"

• Solution: When "Custom Model" is selected, you must provide a model name in the Custom Model option.

Error: "Unsupported image format"

• Solution: Images must be JPEG, PNG, WebP, or GIF. Convert your image to a supported format.

Error: "OpenAI API error: rate limit exceeded"

• Solution: You've exceeded your API rate limit. Wait and retry, or upgrade your OpenAI plan.

Error: "Image file does not exist"

• Solution: Check that the image file path is correct and the file exists at that location.

Unexpected JSON output

• Solution: When using JSON Mode, make sure to explicitly ask for JSON in your prompt (e.g., "Return the data as JSON").

Model produces inconsistent results

• Solution: Lower the temperature (0-0.3) or set a seed value for more deterministic outputs.