Modal Notebooks are a collaborative environment for high-performance interactive computing. They let you explore data, run code, and test ideas in a shared editor. Powered by a GPU-enabled Python kernel that launches in seconds on Modal’s AI infrastructure.
I’m thrilled to share Modal Notebooks: the fastest, most powerful cloud GPU notebook. It has modern real-time collaborative editing and is backed by our AI infrastructure — swap GPUs in seconds.
I started hacking on Jupyter notebooks because they’re incredibly useful, a pillar of research and fast iteration. But it’s too easy to lose your work, the documents are siloed, and few providers offer high-end GPUs. That’s not good DevEx!
We reimagined notebooks from the ground up:
- Each notebook can have up to 8 B200 GPUs and starts in seconds
- Use arbitrary custom images
- Modern features: AI, LSP, semantic highlighting
- Jupyter Widgets synced across users
- A beautiful, sleek UI
In our short beta period last month, 5000+ accounts have already started using Modal Notebooks to run 200,000+ code cells. One of those is Suno (https://www.producthunt.com/prod...), where researchers share models and experiments across the company.
More info in our product announcement: https://modal.com/blog/notebooks
Today, thousands of developers are already using Modal Notebooks to research and build. We hope they allow people to work together more effectively and ultimately bring simpler, better technology into the world.
Happy to answer any questions!
I can’t say enough good things about Modal: deploying data apps and ML models here is seamless, and ditching the hassle of owning infrastructure lets me focus fully on building, not maintaining.
i've been hoping for notebooks for a long time. so glad to see them going live. nice work, team!
This is the truly the next generation cloud infrastructure company. With ever more increasing AI models we need such companies to support for building and inference.
This is really great! I have used similar tools before to run Jupyter notebooks that needed GPU power, but I found them very frustrating since it often took several minutes (sometimes even 15+ min) to be allocated the GPUs, and even if some were faster, if I have needed to work with others and share the project, I have had to send files around. No more such problems! Congratulations on this fantastic product!
