Create New GPU Container
Create New GPU Container
Updated on 22 May 2025

Info Icon
Notice: Each tenant can only have a maximum of 10 containers. If you have reached this limit, please delete unused container to create a new one.

Step 1: In the AI Factory Portal, navigate to “GPU Container". From the list view, click button “Create New Container”

Alt text

Step 2: Give your container a name

Alt text

Step 3: Select GPU Instance

Currently, GPU Container only support card NVIDIA H100. Alt text

Step 4: Choose Template

We highly recommend our customers use built-in templates for faster deployment. Currently, GPU Container provides 5 templates: Ollama, Ollama WebUI, vLLM, Code Server, Jupyter Notebook.

  • Built-in templates: user clicks on the dropdown icon at the end of template card.
  • Custom template: in case users want to use their own image or use another image version, they can use feature “Custom Template”
    • Click button “Custom Template”, a modal will be opened for user to input the new image information
    • Click button “Save Changes” to save new image information

Step 5 (Optional): Configure the Advance Settings based on demands.

This section includes:

  • HTTP Port: this information is required
  • Persistent Disk: specify the amount of storage that users need to store training weights, models, etc.
  • Environment Variables: key-value pairs injected into the container at runtime.
  • Startup Command: command and arguments to run at the start of container.

Step 6: Click “Create New Container” to create.

Alt text

Step 7: In case your balance is insufficient to create a new container, please navigate to Billing to add more credit, details instruction can be find here