​
Overview

Highlight allows you to choose from various AI models to suit different tasks:

  • Need to analyze large PDF files? Use Gemini Pro
  • Looking for fast replies? Try GPT4o
  • The model selection page provides detailed information about each available model

​
Setting Your Default Model

You can set a model as your “daily driver” for regular use. This model will be used by default for your conversations, but you can easily switch to other models on the fly using the mentions menu.

​
PRO Models Access Options

If you have Highlight PRO, you can access flagship models without requiring external subscriptions. Highlight adds new models as soon as they become available. Don’t want to subscribe? You can bring your own API key to access PRO models at public API costs (though this may cost more than a Highlight Pro subscription).

​
Adding Custom Models

For those wanting to test brand new models, Highlight allows you to add custom models with your preferred provider:

  1. Select a provider (e.g., OpenAI) from the dropdown (You can also set Base URL by selecting Custom provider)
  2. Enter the model name (e.g., gpt-4o)
  3. Add your API key
  4. Complete optional fields like display name and description
  5. Click “Add Provider”

You will not be able to edit your API key after you submit your model!

​
Model Capabilities

When adding a custom model, selecting the correct capabilities is essential for proper functionality:

  • Streaming: Enables real-time responses as they’re generated
  • Function Calling: Allows the model to use tools and functions
  • Parallel Function Calling: Supports calling multiple functions simultaneously
  • Response Schema: Supports structured outputs in specific formats
  • Vision: Enables the model to process and understand images
  • Prompt Caching: Stores prompts for improved performance
  • System Message: Supports system-level instructions
  • Developer Message: Supports developer-specific instructions
  • Tool Choice: Allows the model to select appropriate tools
  • Thinking: Supports extended reasoning capabilities
  • Temperature: Allows adjusting randomness in model responses

Select only the capabilities that your specific model actually supports to ensure optimal performance within Highlight.

​
FAQ

  • How to set a ‘default’ model?

    Look for the “Make default” button next to your preferred model in the models page, or check for the “Default” label in the models dropdown menu.

  • How to add a custom model?

    Click “Add A Custom Model” button in the models page, then fill out the required provider information, model name, API key, and select appropriate capabilities.

  • How can I use my own key?

    Navigate to the model settings page where you can enter your own API keys for different providers (OpenAI, Anthropic, Google). Input your key in the appropriate field and click “Save” to start using the model with your own credentials.