Open Interpreter, an open-source initiative, allows LLMs to run code on your computer for a variety of tasks. It not only matches but also goes beyond the capabilities that Windows 11 Copilot promises for the future.

! Observe the striking illustration yourself, see “Comparison of Normalised Stock Prices for AAPL (Apple) and META” below.

Interactions with Open Interpreter go via the terminal. A more intuitive desktop application is currently under development. Early access to this application is available for interested users through this link.

Take aways:

  • Features:
    • Provides a ChatGPT-like interface for natural language interaction (surely voice commands will follow).
    • Allows control over a Chrome browser, data analysis, and more.
    • Supports multiple languages like Python, JavaScript, Shell, etc.
  • Safety Measures: The software asks for user approval before executing any code to ensure safety.
  • Flexibility: Unlike OpenAI’s hosted service, Open Interpreter is not restricted by internet access, time, or file size limits. It can utilize any package or library.
  • Commands and Usage: Various commands are available for interactive chats, programmatic chats, and more. It also supports Azure deployments and debug modes.
  • Customization: Users can customize system messages and behaviors using a .env file.
  • Package and Library Support: Can utilize any package or library available in your local environment.
  • Code Llama Support: You can run the interpreter in local mode to use Code Llama, offering additional flexibility.
  • General-Purpose Capabilities:
    • Create and edit photos, videos, PDFs, etc.
    • Control computer settings like light/dark them, open a Chrome browser
    • Plot, clean, and analyze large datasets.
  • License: The project is licensed under the MIT License and is not affiliated with OpenAI.
  • Community: The project has a strong community with guidelines for contributing.

Examples:

  • Simple change to dark mode of computer:
  • A more complicated task = “Plot AAPL (Apple) and META’s normalized stock prices”:

Several confirmation steps go in between the final result; however, watch here how it corrects its own mistakes:

Endresult:

What it did, recap:

GitHub – KillianLucas/open-interpreter: OpenAI’s Code Interpreter in your terminal, running locally
OpenAI’s Code Interpreter in your terminal, running locally – GitHub – KillianLucas/open-interpreter: OpenAI’s Code Interpreter in your terminal, running locally