Skip to main content

Large Language Model

Execute language tasks using advanced AI models from multiple providers with comprehensive multimodal support.

What is the Large Language Model?​

The Large Language Model (LLM) node is a powerful and versatile component that leverages state-of-the-art language models from leading AI providers. This advanced node supports text generation, multimodal processing, reasoning tasks, and streaming outputs with extensive configuration options for optimal performance across diverse use cases.

How to use it?​

To effectively utilize the LLM node for your AI workflows:

  1. Select Your Model:

    Choose from the extensive collection of supported models:

    OpenAI Models:

    • GPT-5 Series: GPT-5, GPT-5 Mini, GPT-5 Nano
    • GPT-4 Series: GPT-4o, GPT-4o Mini, GPT-4 Turbo 128K, GPT-4 8K, GPT-4.1, GPT-4.1 Mini, GPT-4.1 Nano
    • o-Series: o1, o3, o3-mini, o4-mini (Advanced reasoning models)

    Anthropic Claude Models:

    • Claude 3 Series: Haiku 3, Haiku 3.5
    • Claude 4 Series: Sonnet 4, Opus 4.1, Sonnet 4.5

    Google AI Models:

    • Gemini 2.5: Flash, Pro

    Meta Llama Models:

    • Llama 3 Series: 8B, 70B Instruct
    • Llama 3.1 Series: 8B, 70B, 405B Instruct
    • Llama 3.2 Series: 1B, 3B Instruct
    • Llama 3.3: 70B Instruct

    Amazon Models:

    • Nova Series: Micro, Lite, Pro, Premier

    Writer Models:

    • Palmyra: X4, X5

  2. Configure Authentication:

    Set up appropriate credentials based on your selected model:

    • OpenAI Models: OpenAI API key and organization ID
    • Google AI Models: Generic API key credentials
    • AWS Bedrock Models: AWS credentials with Bedrock permissions
    • Managed Credentials: Use Nocodo AI's managed credential system

  3. Select Region:

    Choose the appropriate region for your model:

    • OpenAI: No region selection required
    • Google: Select from available GCP regions
    • AWS Bedrock: Select from available AWS regions (us-east-1, eu-west-2, etc.)
    • Region-specific Models: Some models are only available in specific regions

  4. Configure Model Parameters:

    Temperature (0.0 - 2.0):

    • Controls randomness and creativity in responses
    • Lower values (0.1-0.3) for factual, consistent outputs
    • Higher values (0.7-1.0) for creative, diverse responses
    • Default: 0.7

    Top P (0.0 - 1.0):

    • Controls diversity by limiting token selection
    • Lower values for more focused responses
    • Higher values for more varied outputs
    • Default: 0.9

    Response Length:

    • Maximum number of tokens to generate
    • Configure based on your expected output size
    • Default: 1024

    Max Allowed Steps (for tool support):

    • Maximum steps with tool usage
    • Default: 5

  5. Provide Input Content:

    Prompt Text (Required):

    • Connect a string input containing your main prompt
    • Can be dynamic content from previous nodes
    • Supports complex instructions and context

    System Prompt (Optional):

    • Define model behavior and role instructions
    • Set consistent guidelines across interactions
    • Establish context and constraints

    Image Input (Optional - Multimodal Models):

    • Upload images for visual analysis tasks
    • Format support: JPEG, PNG, WebP
    • Use cases: image description, analysis, OCR, visual Q&A

  6. Configure Output Type:

    Text Output:

    • Complete response delivered after processing
    • Best for final results and complete responses
    • Suitable for most standard use cases

    Streaming Output:

    • Real-time token delivery as model generates
    • Enables responsive user interfaces
    • Ideal for chat applications and live interactions

Model Selection Guide​

Task-Specific Recommendations​

  • Creative Writing: GPT-4o, Claude 4 Sonnet, Gemini 2.5 Pro
  • Code Generation: GPT-4 Turbo, o3, Claude 3.7 Sonnet
  • Analysis & Reasoning: o1, o3, Claude 4 Opus
  • Speed & Efficiency: GPT-4o Mini, Nova Lite, Llama 3.2
  • Multimodal Tasks: GPT-5, Claude 3.5+, Gemini 2.5, Nova Pro
  • Cost Optimization: Nova Micro, Llama 3.1, GPT-4o Mini

Best Practices​

Prompt Engineering​

Structure Guidelines:

  • Use clear, specific instructions
  • Provide relevant context and examples
  • Define output format requirements
  • Include constraints and limitations

System Prompts:

  • Establish consistent model behavior
  • Define role and expertise areas
  • Set response tone and style
  • Include safety and ethical guidelines

Authentication & Setup​

OpenAI Models:

  • OpenAI API Keys
  • Organization ID configuration
  • Usage monitoring and billing management

AWS Bedrock Models:

Google AI Models:

Troubleshooting​

Common Issues​

Authentication Errors:

  • Verify API keys and credentials
  • Check service account permissions
  • Validate region access rights
  • Review billing and quota status

Performance Issues:

  • Optimize prompt length and complexity
  • Adjust model parameters appropriately
  • Monitor regional latency patterns
  • Implement proper timeout handling

Quality Issues:

  • Refine prompt engineering approach
  • Experiment with different models
  • Adjust temperature and top-p settings
  • Implement output validation and filtering