Jupyter Book

Overview

Jupyter Book is a publishing tool for turning technical writing and computational work into books for the web and other formats. It is best suited to researchers, educators, and data scientists who want explanations and executable material in the same publication. Its authoring range is a strong match for that work, but the transition to version 2 brings meaningful compatibility trade-offs.

Key features

Authors can bring MyST Markdown and Jupyter notebooks together with cross-references, citations, figures, tables, equations, and executable code blocks. That combination is especially useful when a reader should be able to connect an explanation to the computation behind it. Books can be built as websites, PDFs, or other formats, then published online.

Code execution at build time can generate figures, tables, and values for pages, but it is disabled by default. This leaves authors in control of whether a build runs code; projects that depend on computed output need to enable execution deliberately. Readers can also encounter interactive widgets, run code in the browser through a remote kernel, or launch Binder and JupyterHub environments. Jupyter Book 2 does not yet provide execution feature parity, so teams relying heavily on these workflows should check whether the current capabilities meet their needs.

Plugins can add custom directives, roles, and content transformations, and can be distributed as files, npm packages, or release artifacts. Installation is possible through PyPI, conda-forge, or npm, with methods including pip, mamba or conda, pipx, uv, and pixi. The jupyter book start command serves a local preview and refreshes it as content changes, a useful feedback loop while writing.

Publishing options include GitHub Pages, Read the Docs, GitLab Pages, Netlify, Curvenote, and custom web servers. The project is developed by the Executable Books community, which welcomes feedback and contributions. Discord is the route for quick questions, GitHub Discussions for detailed ones, and most Jupyter Book 2 bug reports go to the mystmd repository.

Pricing

Jupyter Book is open source and free: its Jupyter Book plan costs nothing and can be installed from PyPI, conda-forge, or npm. There is no paid tier or free trial to weigh, and custom domains are supported. Documentation versioning is not supported, so projects that need readers to browse multiple documentation versions will need another approach.

Platforms

Jupyter Book supports Linux, macOS, and Windows, as well as web and self-hosted use. That breadth gives teams flexibility to write and build in their own environment and choose where to publish.

Who it's for

Choose Jupyter Book when a publication needs to combine technical prose, notebook content, equations, and computation, particularly for research, teaching, or data-science material. Its multiple output formats and hosting choices suit teams publishing a book in more than one way. It is a less straightforward fit for projects built around Sphinx-specific customizations: Jupyter Book 2 is based on the MyST Document Engine rather than Sphinx, does not support custom Sphinx extensions, and cannot automatically upgrade books that directly use Sphinx's conf.py.

Version choice matters. Jupyter Book 1 is mostly in maintenance mode, with no active development planned for its tools and limited commitments to review community contributions. Version 2 is a rewrite and the project says it still has substantial improvements to make. Its FAQ also identifies editable Thebe cells, Sphinx autodoc-style API documentation, and exporting notebook outputs as MyST as unavailable. Teams migrating established books should assess those gaps before committing.

Pros and cons

Pros

• Prose and computational content fit together. MyST Markdown, notebooks, code blocks, citations, and equations support books that need both narrative and working material.
• Flexible publishing and deployment. Website and PDF output, several hosting options, and custom domains give authors room to choose how readers access a book.
• Free and extensible. The open-source project has plugin mechanisms and community contribution channels without a paid plan to budget for.

Cons

• Version 2 migration can require rework. Custom Sphinx extensions are incompatible, and direct use of conf.py prevents automatic upgrades.
• Some workflows are not yet available in version 2. Execution parity, editable Thebe cells, autodoc-style API documentation, and notebook-output export as MyST remain gaps.
• Version 1 is not an active-development path. Its maintenance status may deter teams that need ongoing development of its tools.
• No documentation versioning. Teams that must preserve and expose multiple documentation versions should account for that limitation.

Alternatives

For a paid-feature set with a defined free allowance, consider Fern: its free plan costs nothing and caps usage at 10 team members, 1,000 pages, 250 AI credits per month, 2,000 page generations per month, and one site. Pick it over Jupyter Book if those managed limits suit your team better than an open-source publishing workflow.

Mintlify offers a free Starter plan at no cost with five editor seats, custom domains, a web editor, authentication, an MCP server, and an API playground; it also offers a free trial. It is worth considering when that editor-and-platform package is a better fit than notebook-centered book authoring.

Sphinx is a free, BSD 2-Clause-licensed option that can be installed locally or self-hosted. It is the more direct choice for teams whose existing documentation depends on Sphinx and its extensions.

VitePress is a free, MIT-licensed static site generator that requires Node.js 22 or higher. DocFX is a free open-source static site generator requiring .NET SDK 8.0 or higher. Dokiel is a free open-source Scenari publishing chain available as a desktop or SCENARIchain installation. These are alternatives to compare when their stated generator or publishing-chain requirements align with your project.

Docsie offers a free knowledge-base plan at no cost with 25,000 AI credits, a custom domain, search, versioning, unlimited viewers, no credit card requirement, and no expiry; it also has a free trial. It may suit teams seeking a knowledge base with versioning rather than a computational book workflow.

Dr.Explain has a free restricted-license plan capped at 20 topics and a Business Temporary subscription at $290 a year for five users. It is another option to compare if those plan terms suit your documentation needs.

Browse Technical Documentation Software for more options.

Verdict

Jupyter Book is a strong choice for researchers, educators, and data scientists publishing material where narrative and computation belong together. Its free, extensible workflow and broad output choices are compelling; the main reason to look elsewhere is a project tied to Sphinx extensions or a version 2 capability that is still missing. For established Sphinx books, verify migration needs before choosing a version or rebuilding around it.