Local Jupyter

Start a local Jupyter server in one click and use it as a kernel — no manual setup, no terminal commands. It joins the Kernels stack and is registered as your default kernel automatically.

What you can do

• Start a token-authenticated local Jupyter server from the Kernels "+" menu.
• Pick CPU or GPU when you start it — GPU is offered only when an NVIDIA CUDA or Apple Metal device is detected.
• Have it set as your default kernel endpoint automatically, ready for notebooks and flows.
• See running status and stop the server when you're done.

Getting started

1. Open the Kernels view and choose ＋ Add Kernel → Start Local Jupyter.
2. Pick CPU or GPU.
3. The server starts on your machine and becomes the default kernel.

The server shuts down when you stop it or quit the app.

Requirements

• Desktop app only.
• A Python environment. The setup card creates a workspace .venv and installs jupyter-server and ipykernel into it for you, and the server always runs in the environment shown in Python Environments.
• Choosing GPU sets device visibility and labels the kernel; actual GPU compute still needs GPU-enabled Python packages (such as CUDA-enabled PyTorch or tensorflow-metal) in your environment.