Choosing marimo#

How is marimo different from Jupyter?#

marimo is a reinvention of the Python notebook as a reproducible, interactive, and shareable Python program that can be executed as scripts or deployed as interactive web apps.

Consistent state. In marimo, your notebook code, outputs, and program state are guaranteed to be consistent. Run a cell and marimo reacts by automatically running the cells that reference its variables. Delete a cell and marimo scrubs its variables from program memory, eliminating hidden state.

Built-in interactivity. marimo also comes with UI elements like sliders, a dataframe transformer, and interactive plots that are automatically synchronized with Python. Interact with an element and the cells that use it are automatically re-run with its latest value.

Pure Python programs. Unlike Jupyter notebooks, marimo notebooks are stored as pure Python files that can be executed as scripts, deployed as interactive web apps, and versioned easily with git.

What problems does marimo solve?#

marimo solves problems in reproducibility, maintainability, interactivity, reusability, and shareability of notebooks.

Reproducibility. In Jupyter notebooks, the code you see doesn’t necessarily match the outputs on the page or the program state. If you delete a cell, its variables stay in memory, which other cells may still reference; users can execute cells in arbitrary order. This leads to widespread reproducibility issues. One study analyzed 1 million Jupyter notebooks and found that 36% of them weren’t reproducible.

In contrast, marimo guarantees that your code, outputs, and program state are consistent, eliminating hidden state and making your notebook reproducible. marimo achieves this by intelligently analyzing your code and understanding the relationships between cells, and automatically re-running cells as needed.

Maintainability. marimo notebooks are stored as pure Python programs (.py files). This lets you version them with git; in contrast, Jupyter notebooks are stored as JSON and require extra steps to version.

Interactivity. marimo notebooks come with UI elements that are automatically synchronized with Python (like sliders, dropdowns); eg, scrub a slider and all cells that reference it are automatically re-run with the new value. This is difficult to get working in Jupyter notebooks.

Reusability. marimo notebooks can be executed as Python scripts from the command-line (since they’re stored as .py files). In contrast, this requires extra steps to do for Jupyter, such as copying and pasting the code out or using external frameworks. In the future, we’ll also let you import symbols (functions, classes) defined in a marimo notebook into other Python programs/notebooks, something you can’t easily do with Jupyter.

Shareability. Every marimo notebook can double as an interactive web app, complete with UI elements, which you can serve using the marimo run command. This isn’t possible in Jupyter without substantial extra effort.

To learn more about problems with traditional notebooks, see these references [1] [2].

How is marimo.ui different from Jupyter widgets?#

Unlike Jupyter widgets, marimo’s interactive elements are automatically synchronized with the Python kernel: no callbacks, no observers, no manually re-running cells.

Using marimo#

Is marimo a notebook or a library?#

marimo is both a notebook and a library.

  • Create marimo notebooks with the editor that opens in your browser when you run marimo edit.

  • Use the marimo library (import marimo as mo) in marimo notebooks. Write markdown with mo.md(...), create stateful interactive elements with mo.ui (mo.ui.slider(...)), and more. See the docs for an API reference.

What’s the difference between a marimo notebook and a marimo app?#

marimo programs are notebooks, apps, or both, depending on how you use them.

There are two ways to interact with a marimo program:

  1. open it as a computational notebook with marimo edit

  2. run it as an interactive app with marimo run

All marimo programs start as notebooks, since they are created with marimo edit. Because marimo notebooks are reactive and have built-in interactive elements, many can easily be made into useful and beautiful apps by simply hiding the notebook code: this is what marimo run does.

Not every notebook needs to be run as an app — marimo notebooks are useful in and of themselves for rapidly exploring data and doing reproducible science. And not every app is improved by interacting with the notebook. In some settings, such as collaborative research, education, and technical presentations, going back and forth between the notebook view and app view (which you can do from marimo edit) can be useful!

How does marimo know what cells to run?#

marimo reads each cell once to determine what global names it defines and what global names it reads. When a cell is run, marimo runs all other cells that read any of the global names it defines. A global name can refer to a variable, class, function, or import.

In other words, marimo uses static analysis to make a dataflow graph out of your cells. Each cell is a node in the graph across which global variables “flow”. Whenever a cell is run, either because you changed its code or interacted with a UI element it reads, all its descendants run in turn.

How do I prevent automatic execution from running expensive cells?#

Reactive (automatic) execution ensures your code and outputs are always in sync, improving reproducibility by eliminating hidden state and out-of-order execution; marimo also takes care to run only the minimal set of cells needed to keep your notebook up to date. But when some cells take a long time to run, it’s understandable to be concerned that automatic execution will kick off expensive cells before you’re ready to run them.

Here are some tips to avoid accidental execution of expensive cells:

  • Disable expensive cells. When a cell is disabled, it and its descendants are blocked from running.

  • Use Python’s functools.cache to cache expensive intermediate computations (see our best practices guide).

  • Wrap UI elements in a form.

  • Use mo.stop to conditionally stop execution of a cell and its descendants.

How do I use sliders and other interactive elements?#

Interactive UI elements like sliders are available in marimo.ui.

  • Assign the UI element to a global variable (slider = mo.ui.slider(0, 100))

  • Include it in the last expression of a cell to display it (slider or mo.md(f"Choose a value: {slider}"))

  • Read its current value in another cell via its value attribute (slider.value)

When a UI element bound to a global variable is interacted with, all cells referencing the global variable are run automatically.

If you have many UI elements or don’t know the elements you’ll create until runtime, use marimo.ui.array and marimo.ui.dictionary to create UI elements that wrap other UI elements (sliders = mo.ui.array([slider(1, 100) for _ in range(n_sliders)])).

All this and more is explained in the UI tutorial. Run it with

marimo tutorial ui

at the command line.

How do I add a submit button to UI elements?#

Use the form method to add a submit button to a UI element. For example,

form = marimo.ui.text_area().form()

When wrapped in a form, the text area’s value will only be sent to Python when you click the submit button. Access the last submitted value of the text area with form.value.

How do I write markdown?#

Import marimo (as mo) in a notebook, and use the mo.md function. Learn more in the outputs guide or by running marimo tutorial markdown.

How do I display plots?#

Include plots in the last expression of a cell to display them, just like all other outputs. If you’re using matplotlib, you can display the Figure object (get the current figure with plt.gcf()). For examples, run the plots tutorial:

marimo tutorial plots

Also see the plotting API reference.

How do I prevent matplotlib plots from being cut off?#

If your legend or axes labels are cut off, try calling plt.tight_layout() before outputting your plot:

import matplotlib.pyplot as plt

plt.plot([-8, 8])
plt.ylabel("my variable")

How do I display interactive matplotlib plots?#

Use marimo.mpl.interactive.

fig, ax = plt.subplots()
ax.plot([1, 2])

How do I display objects in rows and columns?#

Use marimo.hstack and marimo.vstack. See the layout tutorial for details:

marimo tutorial layout

How do I create an output with a dynamic number of UI elements?#

Use mo.ui.array, mo.ui.dictionary, or mo.ui.batch to create a UI element that wraps a dynamic number of other UI elements.

If you need custom formatting, use mo.ui.batch, otherwise use mo.ui.array or mo.ui.dictionary.

For usage examples, see the recipes for grouping UI elements together.

How do I restart a notebook?#

To clear all program memory and restart the notebook from scratch, open the notebook menu in the top right and click “Restart kernel”.

How do I reload modules?#

To reload modules, use importlib.reload():

import mymodule
import importlib

Running this cell will reload mymodule with your new edits and automatically re-run any cells using mymodule.

Why aren’t my on_change/on_click handlers being called?#

A UI Element’s on_change (or for buttons, on_click) handlers are only called if the element is bound to a global variable. For example, this won’t work

mo.vstack([mo.ui.button(on_change=lambda _: print('I was called")) for _ in range(10)])

In such cases (when you want to output a dynamic number of UI elements), you need to use mo.ui.array, mo.ui.dictionary, or mo.ui.batch.

See the recipes for grouping UI elements together for example code.

How does marimo treat type annotations?#

Type annotations are registered as references of a cell, unless they are explicitly written as strings. This helps ensure correctness of code that depends on type annotations at runtime (e.g., Pydantic), while still providing a way to omit annotations from affecting dataflow graph.

For example, in

x: A = ...

A is treated as a reference, used in determining the dataflow graph, but in

x: "A" = ...

A isn’t made a reference.

For Python 3.12+, marimo additionally implements annotation scoping.

How do I use dotenv?#

The package dotenv’s loadenv() function does not work out-of-the box in marimo. Instead, use dotenv.load_dotenv(dotenv.find_dotenv(usecwd=True)).

What packages can I use?#

You can use any Python package. marimo cells run arbitrary Python code.

How do I use marimo on a remote server?#

Use SSH port-forwarding to run the marimo kernel on a remote server and connect to it from a browser on your local machine. Make sure to pass the --headless flag when starting marimo on remote:

marimo edit notebook.py --headless

You may also want to set a custom host and port:

marimo edit notebook.py --headless --host --port 8080

How do I deploy apps?#

Use the marimo CLI’s run command to serve a notebook as an app:

marimo run notebook.py

If you are running marimo inside a Docker container, you may want to run under a different host and port:

marimo run notebook.py --host --port 8080

Is marimo free?#