Ollama

Ollama is an easy way to get local language models running on your computer through a command-line interface.

To run Ollama with Open interpreter:

1. Download Ollama for your platform from here.

2. Open the installed Ollama application, and go through the setup, which will require your password.

3. Now you are ready to download a model. You can view all available models here. To download a model, run:

ollama run <model-name>

4. It will likely take a while to download, but once it does, we are ready to use it with Open Interpreter. You can either run interpreter --local to set it up interactively in the terminal, or do it manually:

interpreter --model ollama/<model-name>

from interpreter import interpreter

interpreter.offline = True # Disables online features like Open Procedures
interpreter.llm.model = "ollama_chat/<model-name>"
interpreter.llm.api_base = "http://localhost:11434"

interpreter.chat()

For any future runs with Ollama, ensure that the Ollama server is running. If using the desktop application, you can check to see if the Ollama menu bar item is active.