What is VoltageGPU API?

VoltageGPU provides two APIs: a GPU Pods API for deploying and managing GPU instances, and an AI Inference API for running 140+ AI models including DeepSeek-R1, Llama 3, and FLUX.

Is VoltageGPU API OpenAI compatible?

Yes, the VoltageGPU AI Inference API is fully OpenAI-compatible. You can use existing OpenAI SDKs by simply changing the base URL to https://api.voltagegpu.com/v1

What AI models are available?

VoltageGPU offers 140+ AI models including DeepSeek-R1, DeepSeek-V3, Llama 3.3 70B, Qwen 2.5, Mixtral, FLUX.1, Stable Diffusion, and many more for chat, embeddings, image and video generation.

Documentation

Templates

Templates in VoltageGPU are configuration blueprints that define how GPU-enabled Docker containers are created and deployed. They provide a standardized way to specify all necessary parameters for container creation.

Overview

Each template contains essential Docker configuration parameters that determine how the GPU environment will be set up, including the base image, environment variables, storage requirements, and network settings. Templates streamline the process of launching GPU environments by providing a reusable configuration that can be easily deployed.

How It Works

Template Structure

A template in VoltageGPU consists of several required components:

Docker Image Configuration

Docker image name
Image tag version
These determine the base environment for your GPU container

Environment Variables

Custom environment variables needed for your applications
System configuration parameters
Runtime settings

Volume Configuration

Storage volume specifications
Mount points for persistent data
Access permissions

Network Settings

Port exposure configurations
Which ports should be accessible
Network interface settings

Startup Commands

Container initialization commands
Service startup parameters
Runtime configurations

Template Types

Official Templates

Maintained by the VoltageGPU team. Verified and trusted to be secure and functional. Includes PyTorch, TensorFlow, JAX, and more.

Custom Templates

Created by users for specific use cases. Can be public (shared with community) or private (only visible to you).

Public Templates

Community-contributed templates available to all users. Great for discovering new configurations and workflows.

Private Templates

Your personal templates visible only to your account. Perfect for proprietary configurations and internal tools.

Template Verification

Why Verification Matters

Custom templates in VoltageGPU are user-defined configurations for GPU environments. While this flexibility is powerful, it's crucial to ensure these templates work correctly before deployment. VoltageGPU implements a robust template verification system that validates all custom templates before they can be used in production.

Verification Process

The verification process consists of four key steps:

1. Test Deployment

Creates a container on dedicated verification servers using exact template configurations.

2. Container Readiness Check

Monitors container startup process and tracks initialization of required services.

3. SSH Connection Validation

Verifies SSH accessibility, confirms proper credential handling, and tests connection stability.

4. Error Reporting

Provides detailed error logs if verification fails with troubleshooting guidance.

Verification Prerequisites

Your container must remain running continuously (no immediate exits)
Docker image should be publicly accessible

Test Your Container Locally

To verify if your container can pass verification:

# Run your container
docker run [-e <env-var>=<value>] -d <your-docker-image>:<image-tag> <your-startup-command>

# Check if container keeps running
docker ps

💡 Tip

If your container exits immediately, check your startup command and ensure any required services are configured to run in the foreground.

Popular Official Templates

Template	Base Image	CUDA Version	Best For
PyTorch (CUDA)	daturaal/pytorch	12.8.0	Deep learning, Computer vision
TensorFlow (CUDA)	tensorflow/tensorflow	12.4.0	Production ML, Keras workflows
JAX (CUDA)	jax/jax	12.6.0	Research, Transformers, TPU-like
Ubuntu Base	nvidia/cuda	12.8.0	Custom setups, Flexibility
Jupyter Lab	jupyter/datascience	12.4.0	Interactive development, Notebooks

Creating Custom Templates

Step-by-Step Guide

1. Navigate to Templates

Go to Dashboard → Templates → Create Template

2. Configure Basic Settings

Template Name: A descriptive name for your template
Description: What this template is for
Visibility: Public or Private

3. Docker Configuration

Image: your-registry/your-image
Tag: latest
Registry: Docker Hub / Private Registry

Environment Variables:
  CUDA_VISIBLE_DEVICES=all
  PYTHONUNBUFFERED=1
  
Startup Command: /bin/bash -c "service ssh start && tail -f /dev/null"

4. Port Configuration

SSH Port: 22 (required)
Jupyter: 8888 (optional)
TensorBoard: 6006 (optional)
Custom ports as needed

5. Submit for Verification

Click "Create & Verify" to submit your template. The verification process typically takes 2-5 minutes.

Frequently Asked Questions

Can I use private Docker registries?

Yes! Add your Docker credentials in Dashboard → Docker Credentials, then reference your private images in templates.

Why did my template verification fail?

Common reasons: container exits immediately, SSH not configured, image not accessible, or startup command errors. Check the verification logs for details.

Can I edit a template after creation?

Yes, you can edit your custom templates. Changes require re-verification before the updated template can be used.

How do I share my template with others?

Set your template visibility to "Public" when creating or editing. Public templates appear in the community templates list.

Ready to Create Your Template?

Build custom GPU environments tailored to your workflow.

Create Template Browse Templates