What is Notebook.link?
Notebook.link reimagines the Jupyter experience for the modern web. It combines the flexibility of traditional notebooks with the power of in-browser computation, enabling instant startup times and zero infrastructure overhead for users.
Notebook.link is a solution to:
Give it a try
You can try Notebook.link using any public GitHub repository URL:
How does it work?
With Notebook.link, you can easily share notebooks within a project.
A project consists of:
- Environment: An isolated workspace where you can install specific package versions. It includes all required dependencies for your project, such as kernels or scientific computing libraries.
- Persistent storage: A dedicated space for your content. In the Jupyter ecosystem, this typically includes notebooks, dashboards, interactive maps, or CAD projects.
Once your project is ready, you can create a shareable link to give access to your work to coworkers or collaborators.
Who is it for?
Whether you're a data scientist, educator, or researcher, Notebook.link offers seamless AI integration and the freedom to run notebooks entirely in your browser. It’s more than a notebook — it’s a platform for building, sharing, and publishing live, interactive content without complexity.
Technologies behind
Notebook.link relies on JupyterLite, a web-based distribution of JupyterLab that runs entirely in the browser, using WebAssembly builds of language kernels and interpreters.
Packages are available through emscripten-forge, a general-purpose software distribution based on conda, tailored for WebAssembly and the web browser. It brings to your webpage languages commonly used in scientific computing, such as Python, R, C++, and more. This is made possible by several JupyterLite kernels : xeus-python, xeus-cpp, xeus-r, xeus-octave, and more.
In addition, packages for Notebook.link environments are hosted on prefix.dev, which provides a fast and reliable package index. Environment solving is performed entirely in the browser using mambajs, a JavaScript-based conda solver powered under the hood by rattler. This setup enables efficient, fully client-side dependency resolution and installation for JupyterLite sessions.