Skip to main content

Model Selection

Configure the AI model providers that power your Autonify agents. This guide shows you how to select and configure different Large Language Models (LLMs) for optimal performance and cost.

VIDEO TUTORIAL⏱️ 1 min

📹 Model Provider Configuration

Learn how to configure different AI model providers to power your Autonify agents with your preferred LLM.

Overview​

Autonify supports multiple AI model providers, allowing you to choose the best LLM for your needs based on:

  • Performance requirements
  • Cost considerations
  • API availability
  • Compliance requirements

Note: Model configuration is only accessible to users with Admin or Owner permissions.

Supported Model Providers​

OpenAI​

  • Default Provider: Pre-configured with OpenAI GPT models
  • Models Available: GPT-4o, GPT-4o-mini, and future models
  • Use Cases: General purpose, high performance tasks

Anthropic​

  • Claude Models: Access to Claude family of models
  • Models Available: Claude 3 Opus, Claude 3 Sonnet, Claude 3 Haiku
  • Use Cases: Complex reasoning, large context windows

Azure OpenAI​

  • Enterprise Integration: Azure-hosted OpenAI models
  • Models Available: Same as OpenAI, hosted on Azure
  • Use Cases: Enterprise deployments, data residency requirements

Configuration Steps​

Accessing Model Settings​

  1. Navigate to the Settings page in Autonify
  2. Click on the System tab (only visible to Admin and Owner users)
  3. The AI Model Configuration card will be displayed

Configuring a Model Provider​

Each model provider requires specific configuration parameters:

OpenAI Configuration​

  1. Model Provider: Select "OpenAI" from the dropdown
  2. Model Name: Enter the model identifier (e.g., gpt-4o, gpt-4o-mini)
    • Placeholder shows: "e.g., gpt-4o, claude-3-opus-20240229"
  3. Temperature: Use the slider to adjust between 0.0 to 1.0
    • The current value is displayed next to the label
  4. API Key: Enter your OpenAI API key (password field)
    • Placeholder: "Enter your API key"
  5. API URL: Enter the API endpoint (optional)

Anthropic Configuration​

  1. Model Provider: Select "Anthropic" from the dropdown
  2. Model Name: Enter the Claude model identifier (e.g., claude-3-opus-20240229)
  3. Temperature: Use the slider to adjust between 0.0 to 1.0
  4. API Key: Enter your Anthropic API key (password field)
  5. API URL: Enter the API endpoint (optional)

Azure OpenAI Configuration​

  1. Model Provider: Select "Azure OpenAI" from the dropdown
  2. Model Name: Enter your Azure deployment name
  3. Temperature: Use the slider to adjust between 0.0 to 1.0
  4. API Key: Enter your Azure OpenAI API key (password field)
  5. API URL: Enter your Azure OpenAI endpoint URL
  6. API Version: Specify the API version (only appears when Azure is selected)
    • Placeholder: "e.g., 2023-05-15"

Testing Your Configuration​

Connection Testing​

Before saving your configuration, test the connection:

  1. Enter your Model Name and API Key (both are required for testing)
  2. Click the Test Connection button (appears as secondary button)
    • Button shows "Testing..." while in progress
    • Button is disabled if Model Name or API Key are missing
  3. Review the test result:
    • Success: Green message box with "Model connection test successful"
    • Failure: Red message box with error details

The test verifies:

  • API key validity
  • Network connectivity
  • Model availability
  • Proper configuration format

Troubleshooting Connection Issues​

Common issues and solutions:

ErrorSolution
Invalid API KeyVerify your API key is correct and active
Connection TimeoutCheck network settings and firewall rules
Model Not FoundEnsure model name matches provider's format
Rate Limit ExceededWait and retry, or upgrade your API plan
"Failed to test model configuration"Check all required fields are filled correctly

Saving and Managing Configuration​

Save Changes​

  1. Configure your desired settings
  2. Click the Save Changes button (primary button on the left)
    • Button shows "Saving..." while in progress
    • Button is disabled during testing or saving
  3. A success toast appears: "Model configuration saved successfully"
  4. Configuration is applied immediately to all agents

Reset to Defaults​

To restore the default configuration (OpenAI with gpt-4o):

  1. Click Reset to Defaults button (outline button on the right)
    • Button is disabled during testing or saving
  2. Default settings are immediately applied:
    • Model Provider: OpenAI
    • Model Name: gpt-4o
    • Temperature: 0.0
  3. A success toast appears: "Model configuration reset to defaults"

Model Selection Best Practices​

Performance Considerations​

  • OpenAI GPT-4o: Best for complex analysis and reasoning
  • OpenAI GPT-4o-mini: Cost-effective for simpler tasks
  • Claude 3 Opus: Excellent for large context windows
  • Azure OpenAI: Optimal for enterprise security requirements

Cost Optimization​

  1. Start with smaller models (e.g., GPT-4o-mini) for initial testing
  2. Monitor usage through your provider's dashboard
  3. Upgrade selectively based on performance needs
  4. Use temperature settings wisely - lower values reduce token usage

Temperature Settings​

The temperature parameter controls response variability:

  • 0.0 - 0.3: Factual, consistent responses (recommended for data tasks)
  • 0.4 - 0.7: Balanced creativity and accuracy
  • 0.8 - 1.0: Creative, varied responses (use sparingly)

Security Considerations​

API Key Management​

  • Store API keys securely
  • Rotate keys regularly
  • Use separate keys for development and production
  • Monitor usage for anomalies

Network Security​

  • Ensure HTTPS connections
  • Configure firewall rules appropriately
  • Use Azure OpenAI for enhanced security controls
  • Consider VPN or private endpoints for sensitive data

Frequently Asked Questions​

Can I switch between providers?​

Yes, you can change providers at any time through the settings. Changes apply immediately to all future agent runs.

Do different agents use different models?​

Currently, all agents use the same configured model. Per-agent model selection is on the roadmap.

How do I estimate costs?​

Check your provider's pricing page:

What happens if my API key expires?​

Agents will fail to run with an authentication error. Update your API key in settings and test the connection.

Next Steps​