Skip to main content

Get Started

In this short documentation, we lay out everything you need to know to get started using our AI technology. As we improve the models and playground, we will continuously bring the newest tweaks, patches, models and functionalities to you.

The Aleph Alpha API currently includes the following three endpoints:

  • complete: generates completions from a prompt
  • evaluate: evaluates the model's likelihood to produce a completion given a prompt
  • embed: embeds a text and returns vectors that can be used for downstream tasks (e.g. semantic similarity) and models (e.g. classifiers).

As of now, only the complete function is available in the playground. All three endpoints may be accessed using Python. The detailed PyPi documentation using our Python client (or any other http request) for these endpoints and their parameters can be found here.

The following instructions and examples provide best practices and additional resources to the usage of our language models.

Prompt and completion examples use the following format:

User provided prompt text Model completion text