​
Overview of Embedding Models

Our application supports integration with several popular embedding model providers, offering you flexibility in choosing the model that best suits your needs. Here’s an overview of the available models:

​
OpenAI Models

​
text-embedding-3-small

OpenAI’s newer, more efficient embedding model. Good balance of performance and efficiency.

​
text-embedding-3-large

OpenAI’s large embedding model. Offers the best performance in the OpenAI lineup.

​
Credentials Setup

To use OpenAI models:
  1. Sign up for an account at OpenAI’s website.
  2. Follow their docs to get an API key.
  3. Use your OpenAI key in UtopikAI

​
Cohere Models

​
embed-english-v3.0

Cohere’s English embedding model. Good performance for English-language tasks.

​
embed-english-light-v3.0

Cohere’s lightweight English embedding model. Faster and more efficient for simpler tasks.

​
Credentials Setup

To use Cohere models:
  1. Create an account on Cohere’s platform.
  2. Follow their guide to gather an API key.
  3. Use your Cohere API key in UtopikAI.

​
Voyage Models

​
voyage-large-2-instruct

Voyage’s large embedding model. High performance with instruction fine-tuning.

​
voyage-light-2-instruct

Voyage’s lightweight embedding model. Good balance of performance and efficiency.

​
Credentials Setup

To use Voyage models:
  1. Sign up for an account at Voyage AI.
  2. Follow their guide to gather an API key.
  3. Use your Voyage AI API key in UtopikAI.

​
Vertex AI (Google) Model

​
gecko

Google’s Gecko embedding model. Powerful and efficient, but requires more setup.

​
Credentials Setup

To use the Gecko model from Vertex AI:
  1. Set up a Google Cloud Platform (GCP) account if you don’t have one.
  2. Create a new project in the Google Cloud Console.
  3. Enable the Vertex AI API for your project.
  4. Create a service account with the necessary permissions for Vertex AI.
  5. Generate a JSON key for the service account.
  6. Use the JSON key file in UtopikAI by uploading your file.

​
Choosing the Right Model

Consider these factors when selecting an embedding model:
  1. Task Complexity: More complex tasks may benefit from larger models like text-embedding-3-large or voyage-large-2-instruct.
  2. Language Specificity: For English-specific tasks, consider Cohere’s models.
  3. Performance vs. Efficiency: Balance the trade-off between model performance and computational efficiency based on your needs.
  4. Integration Complexity: Consider the ease of setup, especially for models like Gecko that require more configuration.

​
Best Practices

  1. Experiment: Test different models with your typical data to compare performance.
  2. Monitor Performance: Keep track of model performance for your specific use cases.
  3. Stay Updated: Regularly check for updates or new model releases from providers.
  4. Security: Always use the secure interface provided by the application to input your API keys or upload credential files.
Remember to review each provider’s documentation for the most up-to-date information on model capabilities and integration details.