The future of Python notebooks is here
Transform data, train models, and run SQL queries with marimo — feels like an AI-native reactive notebook, stored as Git-friendly reproducible Python. Seamlessly run as scripts and apps. All open source.
$ $ Used by developers at
👁 Johns Hopkins
👁 UC Berkeley
👁 Princeton
👁 Columbia
👁 SLAC National Accelerator Laboratory
👁 Mozilla AI
👁 Chan Zuckerberg Initiative
👁 OpenAI
👁 BlackRock
👁 Capital One
👁 DNB
👁 Shopify
👁 Cloudflare
👁 Moderna
👁 Pfizer
👁 Merck
👁 Bunkerhill Health
👁 Sumble
👁 Cradle
👁 Sonos
👁 Owkin
👁 Stanford
👁 Johns Hopkins
👁 UC Berkeley
👁 Princeton
👁 Columbia
👁 SLAC National Accelerator Laboratory
👁 Mozilla AI
👁 Chan Zuckerberg Initiative
👁 OpenAI
👁 BlackRock
👁 Capital One
👁 DNB
👁 Shopify
👁 Cloudflare
👁 Moderna
👁 Pfizer
👁 Merck
👁 Bunkerhill Health
👁 Sumble
👁 Cradle
👁 Sonos
👁 Owkin
There have been few technologies that have been truly transformational for me. marimo is one of them. Today, just about everything you need to know about Sumble lives in marimo.
My many frustrations with Jupyter notebooks are well known. marimo solves all of them, elegantly, while keeping the same sense of a 'notebook' that many of us have always found so appealing.
The most powerful feature is that these notebooks are reactive: if you change the value or code in a cell every other cell that depends on that value will update automatically.
Marimo is creating some really wonderful interactive computing environments that are intuitive, powerful and versatile.
The notebook, reinvented
Code and outputs stay in sync
Run one cell and marimo reacts by running affected cells, eliminating the error-prone chore of managing notebook state. For expensive notebooks, configure marimo to mark outputs as stale instead of autorunning.
Interactive elements
Built-in interactive elements like sliders, dropdowns, dataframe tables, and selectable plots make working with data feel fast, futuristic, and fun.
Python-first, Git-friendly
marimo notebooks are pure Python, stored as .py files Version with Git, run as Python scripts, reuse as modules, manage with uv and PEP 723 metadata, test with PyTest, and integrate with your favorite developer tools.
Reproducible
Notebooks are executed in a deterministic order,with no hidden state — delete a cell and marimo deletes its variables while updating affected cells. Optionally serialize package requirements in notebook files, with marimo managing virtual environments for you.
Share as apps, run as scripts
Share interactive notebooks on the web by exporting to WebAssembly-powered HTML or serving notebooks as web apps with the marimo CLI. Run as scripts locally or in pipelines.
First-class SQL support
Query dataframes and databases with marimo's built-in SQL cells. These integrate with Polars, Pandas, PyArrow, DuckDB, SQLite, Postgres, MySQL, and many other backends.
Made for developers
The marimo editor comes with GitHub Copilot-style autocomplete, hover tooltips, vim keybindings, code formatting, debugging panels, and extensive hotkeys. marimo also ships with a CLI, a library, and a VS Code extension.
Made for agents
Connect any AI agent to running marimo notebook sessions with marimo pair, and code in an AI-native editor with copilots, AI assistance, inline edits, and context-aware chat, using any model, hosted or local.
