Thunder Compute provides pre-configured instance templates that set up popular AI applications automatically. These templates streamline the deployment process and come with all necessary dependencies pre-installed.

​
Available Templates

Currently, Thunder Compute offers the following templates:

  • Ollama: Sets up an Ollama server for running open-source LLMs
  • ComfyUI: Configures ComfyUI for AI image generation workflows
  • WebUI Forge: Sets up a WebUI Forge server for running Stable Diffusion WebUI Forge

​
Using Templates

  1. Create an instance with your chosen template:
# For Ollama
tnr create --template ollama

# For ComfyUI
tnr create --template comfy-ui

# For WebUI Forge
tnr create --template webui-forge (--gpu a100)

WebUI Forge comes packaged with Flux-fp8 and Flux-fp4. It’s recommended to use an A100 instance for best performance.

  1. Connect to your instance:
tnr connect 0  # Replace 0 with your instance ID

Port forwarding is handled automatically when you connect to a template instance. You don’t need to specify port forwarding manually with the -t flag.

  1. Start the service:
# For Ollama
start-ollama

# For ComfyUI
start-comfyui

# For WebUI Forge
start-webui-forge

The required ports will be automatically forwarded to your local machine.

​
Template Details

​
Ollama Template

  • Automatically forwards port 11434
  • Access the Ollama API at http://localhost:11434
  • Ready to use with popular Ollama models

​
ComfyUI Template

  • Automatically forwards port 8188
  • Automatically mounts the ComfyUI directory from the instance’s home directory on Mac and Linux
  • Access the ComfyUI interface at http://localhost:8188
  • Includes common ComfyUI nodes and extensions

​
WebUI Forge Template

  • Automatically forwards port 7860
  • Automatically mounts the stable-diffusion-webui-forge directory from the instance’s home directory on Mac and Linux
  • Access the WebUI Forge interface at http://localhost:7860
  • Includes Flux-fp8 and Flux-fp4 models

​
Need Help?

If you encounter any issues with instance templates or need assistance, please contact our support team.

Installing CondaDocker