Best / Data science notebooks
Best open-source Data science notebooks
The best self-hosted notebook split in the c17.io catalog is JupyterHub for a BSD-3-Clause multi-user server (the published Google Colab stack), marimo when you want that Apache-2.0 reactive Python file, JupyterLab when the job is the BSD-3-Clause IDE JupyterHub can spawn, and SurfSense when the job is a research Q&A notebook rather than a kernel. This is not Colab's every GPU SKU. Cited research stays on /best/research-notebooks.
Open the stackStart on Railway
Ranked picks
1
JupyterHubBSD-3-Clause multi-user notebook server — the c17 Google Colab path. /compare/jupyterhub-vs-marimo.
When to pick it: Kernels and users you host.
8,362 stars in the catalog
2
MarimoApache-2.0 reactive Python notebook — /compare/jupyterhub-vs-marimo.
When to pick it: A script-shaped notebook, not a multi-user hub.
23,236 stars in the catalog
3
JupyterLabBSD-3-Clause notebook IDE JupyterHub can spawn. Not a hub by itself.
When to pick it: The editor, not the multi-user proxy.
15,373 stars in the catalog
4
SurfSenseApache-2.0-core research notebook already on /best/research-notebooks. Not a data-science kernel.
When to pick it: Cited answers over sources, not a Colab runtime.
How they compare
| Tool | License | Self-host | Best for |
|---|---|---|---|
| JupyterHub | BSD-3-Clause | Medium | Kernels and users you host. |
| Marimo | Apache License 2.0 | Easy | A script-shaped notebook, not a multi-user hub. |
| JupyterLab | BSD-3-Clause | Medium | The editor, not the multi-user proxy. |
| SurfSense | Apache-2.0 (core); BUSL-1.1 (proprietary directory) | Medium | Cited answers over sources, not a Colab runtime. |
Related pages
FAQ
- JupyterHub or marimo?
- JupyterHub if you want the published Colab path. marimo if you want a reactive Python file — /compare/jupyterhub-vs-marimo.
- Is this NotebookLM?
- Cited research stays on /replace/notebooklm (SurfSense). This page is a data-science kernel.
- Are these free?
- JupyterHub and JupyterLab are BSD-3-Clause. marimo is Apache-2.0. You pay hosting and any GPUs you attach. Template users never pay c17.
- How do I deploy it?
- Open /stacks/jupyterhub-notebooks. Persist notebooks. Create one user and one notebook.
Stack of the Week
Get one stack a week that replaces a SaaS bill.