marimo

In marimo, every cell is aware of its dependencies when a cell's output changes, all cells that reference its variables automatically re-execute, maintaining consistency without requiring manual re-runs.

This reactive execution model eliminates the class of bugs caused by running cells out of order or referencing stale variable values, making marimo notebooks inherently reproducible.

Beyond reactivity, marimo supports SQL cells that query dataframes and databases with syntax highlighting and autocomplete, interactive UI elements (sliders, dropdowns, data tables) that trigger reactive updates when changed, and export to pure Python scripts that run without the marimo runtime.

Notebooks are stored as standard Python files rather than JSON, making them diff-friendly for version control and reviewable as code rather than opaque binary blobs. Marimo can also be deployed as interactive web applications by exposing notebook inputs as UI controls.

Marimo is open-source under the Apache 2.0 license and is designed as a modern replacement for Jupyter in both exploratory data analysis and production reporting workflows.

It integrates with the Python data science stack pandas, polars, NumPy, matplotlib, Plotly, Altair and supports package management via uv, making it easy to create reproducible environments.

The cloud-hosted version provides collaborative notebooks with sharing and commenting features, while the local version runs as a web server accessible via browser on any machine with Python installed.