How to Launch AI Notebook
1. Open AI Notebook
In the Workspace sidebar, click AI Notebook under Services. The notebook list shows existing notebooks for your Workspace, one per region.
2. Create a notebook
Click Create Notebook. The create form opens.
Select a region
| Region | Available GPU |
|---|---|
| Hà Nội (Southeast Asia) | H100 SXM5 |
| Tokyo (Japan) | H200 SXM5 |
The region you select determines which GPU options are available.
Only regions enabled for your Workspace appear here. If a region is missing, contact your Organization Owner.
Select a GPU instance
- In the GPU Instance section, select a flavor card.
- Each card shows the GPU count, VRAM, CPU, RAM, and hourly cost.
- Cards marked Out of stock are not available — select a different flavor.
Charges start when the notebook reaches Running. The pricing panel shows the estimated hourly cost before you confirm.
Configure persistent storage
- In the Persistent Storage section, enter a disk size in GB. Minimum is 50 GB, maximum is 10,000 GB.
- The mount path defaults to
/home/user.
Persistent storage retains your files when the notebook is stopped or re-created. GPU instance data does not persist between sessions.
3. Wait for Running status
After clicking Create Notebook, the notebook card appears with status Creating. The GPU instance is being provisioned — this typically takes 3–5 minutes.
| Status | What it means |
|---|---|
| Creating | GPU instance is being provisioned. Wait 3–5 minutes. |
| Running | Notebook is active. Billing is accumulating. |
| Starting | Notebook is starting after a Stop/Start cycle. Wait 1–2 minutes. |
| Stopped | Notebook is inactive. Billing has stopped. Persistent storage is retained. |
| Failed | Provisioning failed. Click Re-Create to try again. |
If the status does not change to Running after 10 minutes, see Troubleshooting.
4. Access the notebook
When the status is Running, click Access to open JupyterLab in a new tab. If a pop-up blocker prevents it, click Open AI Notebook to enter the Launcher manually.
5. JupyterLab Launcher overview
The Launcher is your starting point inside JupyterLab.
Navigation bar
| Item | What it does |
|---|---|
| File Browser | Manage files and folders — open, rename, delete, organize. |
| Running Kernels / Notebooks | View active kernels and open notebooks. Stop idle ones to free resources. |
| Table of Contents | Jump between headings in long notebooks. |
| GPU Kernel Management | Monitor GPU allocation and usage. Shut down unused kernels. |
| Extensions Manager | Install, enable, or disable extensions (e.g. Git, code formatters). |
Notebook and Console options
Notebook — interactive code + output environment. Supports Markdown, charts, and inline documentation. Resource options: CPU (Free), 1×, 2×, 4×, 8× GPU H100 SXM5.
Console — lightweight command-line interface for quick scripts without rich output. Same resource options.
Other — create standalone Text, Markdown, or Python files.
Resource and pricing panel
Displays available GPU configurations with CPU, RAM, VRAM, and hourly cost. Disconnect and delete unused runtimes to avoid unnecessary charges.