Jupyter Notebook: the free notebook where code, output and explanation share a page

Jupyter Notebook is a BSD-3 web application for interactive computing. You run code in cells, keep the result in the document, and explain it in the same file with prose, equations and charts. It runs on your own machine or server, and the browser is the client.

How it works

A kernel is a separate process that holds your variables and executes what you send it. Each cell is a block of code, and the output stays attached to that cell, which is why a notebook reads like a lab record rather than a script. The file is JSON with the code, the outputs, the prose and the metadata in one place, which is what makes it shareable and also what makes it awkward in git.

When the analysis is done, the document exports to HTML, PDF or slides. That is the part people underestimate: the notebook is both the working tool and the report.

What you need

• Python 3, installed directly or through the jupyter/scipy-notebook image.
• A browser. The interface is web based even when it runs locally.
• Port 8888 open if you run it on a server.
• Somewhere to keep the .ipynb files, ideally in version control once the analysis settles.

What it does well

• Exploring data step by step, seeing the output before deciding the next step.
• Keeping code, charts and the reasoning around them in one document a non-coder can read.
• Any kernel, so the same interface serves Python, R and Julia work.
• Exporting to HTML, PDF or slides without rewriting anything.
• Running the same notebook locally, on a VPS, or inside a container with a pinned environment.

Where it falls short

• Hidden state. Run cells out of order and the document no longer reflects what produced the numbers you are about to present.
• Diffs. Notebook JSON is noisy in git, so reviewing a change means reading a rendered diff or skipping review.
• One document at a time, with no file tree or terminal beside it.
• It is not a production runtime. Scheduling, serving and access control are things you build around a notebook, not features inside it.
• Left unattended on an open port, it is an execution engine anyone can use. Put authentication in front of it.

How to install Jupyter Notebook

Option 1, one-click install: Hostinger deploys Jupyter Notebook as a preconfigured Docker app on a managed server, so there is no command line and nothing to configure. It runs notebooks with live code and visualisations.

Option 2, install it yourself: pip install notebook and run jupyter notebook, or start the Docker image with docker run -p 8888:8888 jupyter/scipy-notebook and open the printed URL.

FAQ

Is Jupyter Notebook free? Yes, BSD-3-Clause, including for commercial work. You can modify and redistribute it.

Is it the same as JupyterLab? No. Both come from Project Jupyter and share kernels and the .ipynb format, but Notebook shows one document while JupyterLab arranges notebooks, consoles and terminals in a workspace.

Can I run it on a server and reach it from a browser? Yes, that is the common VPS setup. Add authentication before you expose the port, because a notebook server runs code by design.

Why do my results change between runs? Cells ran out of order. Restart the kernel and run all, top to bottom, before trusting the output.

Does it work with R and Julia? Yes, through the respective kernels. Install the kernel and it appears in the launcher.

Can I keep notebooks in git without the mess? Keep them, but treat the output as disposable: clear outputs before committing, or keep the code in a module and the notebook as the report.

Last checked: November 2026. Licence read from the repository, and the maintained version lines from the project README.