Manage AI Notebooks
This page explains how to stop, start, and delete an AI Notebook, and how the notebook's status affects billing.
Notebook status lifecycle
A notebook moves through the following states:
| Status | What it means |
|---|---|
| Creating… | The GPU instance is being provisioned. Wait 3–5 minutes. |
| Running | The notebook is active. Billing is accumulating. |
| Starting… | The notebook is starting up after a Stop/Start cycle. Wait 1–2 minutes. |
| Stopped | The notebook is inactive. Billing has stopped. Your persistent storage is retained. |
| Failed | Provisioning failed. Click Re-Create to try again. |
Billing starts the moment the notebook reaches Running. It stops only when you explicitly stop the notebook — not when you close the browser or log out.
Stop a notebook
Stopping a notebook releases the GPU instance and stops billing. Your persistent disk data is retained.
- On the notebook list, find the notebook with status Running.
- Click Stop.
- Confirm in the dialog.
The status changes to Stopped. Billing stops at this point.
Stopping a notebook terminates all running kernels. Save your work in JupyterLab before stopping.
Start a stopped notebook
- On the notebook list, find the notebook with status Stopped.
- Click Start.
The status changes to Starting…, then to Running within 1–2 minutes. Billing resumes when the status reaches Running.
Re-create a failed notebook
If a notebook reaches Failed status:
- On the notebook list, find the notebook showing Failed.
- Click Re-Create.
The portal attempts to provision a new GPU instance. The status returns to Creating….
If the notebook fails repeatedly, see Troubleshooting or contact your Workspace Admin.
Delete a notebook
Deleting a notebook permanently removes the notebook and its persistent storage. This action cannot be undone.
- On the notebook list, click the Access button on the notebook you want to delete. This opens the notebook detail page.
- Click Delete in the top-right corner of the detail page.
- Read the confirmation — deleting stops all running kernels and removes all stored data.
- Click Delete to confirm.
Deletion cannot be undone. Back up any files you need from JupyterLab before deleting.
Understanding billing for notebooks
- Billing starts when the notebook reaches Running.
- Billing stops when you click Stop and the status changes to Stopped.
- Storage cost continues while the notebook exists, even when stopped — persistent disk is billed at a lower rate than the GPU instance.
- Deleting stops all charges including storage.
To check your Workspace's current credit usage, go to Billing in the Workspace sidebar.
Idle behavior and auto-shutdown
AI Notebook automatically disconnects idle sessions to prevent unexpected charges.
| Trigger | Idle threshold | What happens |
|---|---|---|
| GPU Kernel idle | 30 minutes of no code execution | Kernel is automatically disconnected. |
| Lab Environment idle | 1 hour of no interaction (no typing, clicking, or tab switching) | Lab is automatically deleted to free resources. |
If the lab is deleted due to idleness, your saved files are still in persistent storage. Only unsaved work (in-memory variables, unsaved editor buffers) is lost. Save frequently.
What's next
- Troubleshooting — if creation fails or the notebook stays in Creating… too long.
- Creating a GPU Container — for containerized workloads requiring custom environments.