Skip to content

Sources Overview

Sources in OpenGPU are the foundation for publishing and managing your computational tasks on the decentralized network. A source defines the container environment, payment structure, and delivery method for your tasks.

🎯 What is a Source?

A Source is essentially a configuration template that specifies:

  • Container Environments: Public URLs to Docker Compose files defining service infrastructure
  • Payment Settings: Minimum payment requirements and lockup amounts
  • Delivery Method: How task results are delivered and confirmed
  • Metadata: Name, description, and branding information

🏗️ Source Components

Image Environments

Sources support multiple hardware environments through public URLs to Docker Compose files:

  • CPU: Standard CPU-based processing
  • NVIDIA: NVIDIA GPU acceleration
  • AMD: AMD GPU acceleration

Technical Implementation

OpenGPU sources use public URLs pointing to Docker Compose files, not direct Docker image references. This allows for complete infrastructure definition including networking, volumes, and multi-service deployments.

Delivery Methods

Choose how task results are handled:

  • Manual Confirmation: Client manually confirms responses
  • First Response: First valid response is automatically accepted

Payment Structure

Configure economic parameters:

  • Minimum Payment: Base payment for task execution
  • Minimum Lockup: Required stake from providers
  • Expiry Duration: Maximum time for task completion

🚀 Getting Started

Choose your path based on your needs:

🎯 Two Ways to Create Sources

There are two approaches to create sources on OpenGPU:

🚀 Method 1: Pre-built Templates

Use ready-made source templates provided by OpenGPU. These templates offer zero-configuration deployment for popular frameworks and use cases.

Perfect for:

  • AI/ML tasks using popular frameworks
  • Standard data processing workflows
  • Quick prototyping and testing
  • Production-ready configurations

Benefits:

  • ⚡ Setup in minutes
  • 🛡️ Maintained by OpenGPU team
  • 📊 Proven and tested environments
  • ❌ No Docker knowledge required

Get Started: - Source Templates → - Browse available templates - Hugging Face Pipeline → - AI/ML template example

🔧 Method 2: Custom Sources

Build and deploy your own Docker containers with custom logic, dependencies, and configurations. Perfect for specialized workloads and custom services.

Perfect for:

  • Custom algorithms and specialized processing
  • Proprietary models and business logic
  • Specific hardware optimizations
  • Complex multi-service architectures

Benefits:

  • 🚀 Maximum flexibility and control
  • 🎛️ Custom business logic implementation
  • 💪 Specialized hardware optimizations
  • 🔧 Complete environment customization

Get Started:

Which Approach to Choose?

  • Pre-built Templates: Quick setup, no Docker required, proven configurations
  • Custom Sources: Maximum flexibility, custom logic, specialized dependencies

🔄 Source Lifecycle

  1. Create: Define source configuration and metadata
  2. Publish: Deploy source to the blockchain
  3. Use: Reference source when publishing tasks
  4. Update: Modify source parameters as needed

🎯 Next Steps