Installation

Install ipyllm from PyPI into the same Python environment used by IPython:

python -m pip install ipyllm

The project requires Python 3.11 or newer. If you are installing from a local project checkout, use:

python -m pip install .

Configure a model

ipyllm uses the model and provider configuration managed by llm. Configure a provider key, then choose a default model:

llm keys set PROVIDER
llm models
llm models default MODEL

For local Ollama models, the llm-ollama plugin is included. Install and start Ollama, then select an installed model with llm models default MODEL.

Install remote provider support

The optional providers dependency group installs plugins for Anthropic, DeepSeek, Gemini, Grok, Mistral, OpenRouter, and Perplexity. In a checkout of this project, install the group and run IPython in the resulting environment:

uv sync --group providers
uv run ipython

uv run uses the project environment, so you do not need to activate it manually. If you prefer an activated shell, use source .venv/bin/activate on macOS/Linux or .venv\Scripts\activate on Windows, then run ipython.

The group installs these plugins: llm-anthropic, llm-deepseek, llm-gemini, llm-grok, llm-mistral, llm-openrouter, and llm-perplexity. If you installed ipyllm into another environment rather than syncing this checkout, install the plugin package(s) you need into that same environment, for example:

python -m pip install llm-anthropic llm-gemini

Configure credentials for each provider using the key name documented by its plugin. You can store a key with llm keys set PROVIDER, or set the provider's documented environment variable in the shell before starting IPython. For example, llm keys set anthropic stores an Anthropic key; environment variables must be available to the IPython process.

Then list available models and choose one:

llm models
llm models default MODEL

See the llm remote API plugin directory for the supported providers and links to plugin-specific setup instructions. The llm API key guide covers stored keys and environment variables.

Load the extension

Start IPython or a Jupyter IPython kernel from the configured environment and load the extension:

%load_ext ipyllm

To load it automatically each time IPython starts, add this to the profile's ipython_config.py:

c.InteractiveShellApp.extensions = ["ipyllm"]

See the IPython configuration guide for profile locations and configuration details.