Templates
Pulserun offers pre-configured GPU environments so you can start working immediately.
Available Templates
| Template | Description | Min VRAM | Default Ports |
|---|---|---|---|
| PyTorch 2.3 | ML training with CUDA 12.1 | 8 GB | 8888 |
| Jupyter Lab | Interactive notebooks with GPU | 8 GB | 8888 |
| ComfyUI | Stable Diffusion node-based UI | 12 GB | 8188 |
| vLLM | High-throughput LLM inference | 24 GB | 8000 |
| Ollama | Run LLMs locally | 8 GB | 11434 |
| TensorRT-LLM | Optimized LLM inference | 24 GB | 8000 |
| Custom | Your own Docker image | — | Configurable |
Selecting a Template
When launching an instance, choose from the template gallery. Each template shows:
- Description and use case
- Minimum VRAM requirement
- Pre-installed software
- Default exposed ports
Custom Docker Images
Select Custom and provide your Docker image:
{
"template": "custom",
"docker_image": "your-registry/your-image:tag",
"ports": [8080, 8888]
}See Custom Templates for details.