Qudata.ai

Templates

A template is a saved Docker environment configuration for launching an instance.

What is a template

A template describes what to run on a rented server: Docker image, environment variables, ports, SSH/UI access, and startup parameters.

When creating an instance, you select two objects:

  • Template - environment configuration (what to run).
  • Offer - GPU server (where to run).

Template configuration is fixed at instance startup. Later template edits do not affect already running instances.

You can manage templates on the templates page.

Template fields

Basic

FieldDescription
NameTemplate name
DescriptionPurpose and notes
ReadmeExtended markdown documentation

Docker

FieldDescription
Docker ImageImage in repo/image format
Version TagImage tag. [Automatic] lets platform choose by CUDA version
Docker AuthCredentials for private registries
CommandOn-start script (runs on container start)
Environment VariablesKey-value environment variables

Networking and access

FieldDescription
PortsTCP/UDP ports to expose
SSHEnable SSH access (port 22)
Web UIEnable access to container web interface
UI PortPort where web interface runs
UI PathPath to web interface (/, /ui, etc.)

Hardware

FieldDescription
Min DiskMinimum disk size (GB)
Recommended GPUsCompatible GPU models

Run modes

Behavior depends on enabled options in the template.

SSH - connect to container via SSH. Attach your public key on the instance page after startup.

Web UI - platform proxies access to container web interface. Access token is generated automatically.

docker ENTRYPOINT - if neither SSH nor UI is enabled, container starts with original image ENTRYPOINT.

In SSH and Web UI modes, the container entrypoint is modified by platform. Use Command for custom startup logic.

Public and private templates

Private templates are visible only to you. This is the default mode.

Public templates are available to all users in the catalog. Platform verifies and publishes templates such as JupyterLab, Ollama, ComfyUI, and other ready environments.

Platform presets

When creating an instance in quick start, you can pick a ready preset. Platform automatically applies a verified public template:

  • JupyterLab - Jupyter environment with GPU support.
  • Ollama - local LLM runtime.
  • ComfyUI - graph-based image generation UI.

Presets differ from user templates only by being preconfigured and compatibility-tested.

Create a template

  1. Open the templates page and click Create.
  2. Fill in image, tag, ports, and variables.
  3. Enable SSH and/or Web UI if needed.
  4. Save template.

After saving, the template becomes available when creating instances from the marketplace.

You can also create or update templates via API or CLI.

On this page