Tasks¶

Learn how to publish and manage computational tasks on the OpenGPU network. Tasks define what work you want providers to execute and how results should be delivered.

🎯 What is a Task?¶

A Task represents a single computational job that gets executed on the OpenGPU network. It specifies:

Source: Which service/container to use for execution
Input Data: The data and parameters for processing
Payment: How much to pay providers
Expiry: How long the task remains valid
Delivery: How results are confirmed and delivered

🔄 Task Lifecycle¶

graph LR A[Create Task] --> B[Publish to Network] B --> C[Provider Picks Up] C --> D[Execute in Container] D --> E[Submit Response] E --> F[Confirm Result] F --> G[Payment Released]

Create: Define task parameters and input data
Publish: Submit task to the blockchain
Execute: Network providers run your task
Respond: Providers submit results
Confirm: Results are validated and confirmed
Complete: Payment is released to providers

🚀 Quick Start¶

Get started with tasks in just a few steps:

import ogpu.client
from web3 import Web3

# 1. Configure chain
ogpu.client.ChainConfig.set_chain(ogpu.client.ChainId.OGPU_TESTNET)

# 2. Create task
task_info = ogpu.client.TaskInfo(
    source="0x4F1477E0a1DA8340E964D01e32Dff302F3CB203A",
    config=ogpu.client.TaskInput(
        function_name="text2text",
        data={"messages": [{"role": "user", "content": "Hello!"}]}
    ),
    expiryTime=int(time.time()) + 600,
    payment=Web3.to_wei(0.01, "ether")
)

# 3. Publish task
task_address = ogpu.client.publish_task(task_info)
print(f"✅ Task published: {task_address}")

📚 Task Types¶

💬 Text Processing¶

AI Chat: Conversational AI and question answering
Text Analysis: Sentiment, summarization, translation
Content Generation: Articles, code, creative writing

🖼️ Media Processing¶

Image Analysis: Object detection, classification
Image Generation: AI art, photo editing
Video Processing: Transcription, analysis

🧮 Computational Tasks¶

Data Analysis: Statistical processing, modeling
Scientific Computing: Simulations, calculations
Batch Processing: Large-scale data transformation

🛠️ Core Components¶

TaskInfo Structure¶

ogpu.client.TaskInfo(
    source="0x...",              # Source address
    config=TaskInput(...),       # Function and data
    expiryTime=timestamp,        # When task expires
    payment=wei_amount           # Payment in wei
)

Function Mapping¶

Tasks call specific functions in your published sources:

# Your service function
@ogpu.service.expose()
def text2text(input_data: InputData) -> Message:
    # Implementation here
    pass

# Task configuration
config = TaskInput(
    function_name="text2text",   # ← Matches function name
    data={...}                   # ← Matches InputData structure
)

🎯 Task Categories¶

By Complexity¶

Type	Duration	Payment Range	Examples
Simple	< 5 min	0.001-0.005 OGPU	Basic Q&A, simple text
Medium	5-30 min	0.005-0.02 OGPU	AI inference, analysis
Complex	30+ min	0.02-0.1 OGPU	Heavy compute, batch jobs

By Use Case¶

🔍 Research: Data analysis, scientific computing
🎨 Creative: Content generation, design
🤖 AI/ML: Model inference, training
📊 Business: Report generation, automation

🌟 Key Features¶

⚡ Fast Execution¶

Tasks are picked up quickly by available providers

💰 Competitive Pricing¶

Pay only for successful task completion

🔒 Secure & Trustless¶

Blockchain-based payments and verification

🌍 Global Network¶

Access to worldwide distributed computing

📈 Scalable¶

Handle single tasks or large batch operations

🎯 Getting Started¶

Choose your path:

🚀 Quick Publishing¶

Jump straight to creating tasks - Publishing Tasks - Step-by-step publishing guide

⚙️ Detailed Configuration¶

Learn advanced task configuration - Task Configuration - Complete configuration guide

📊 Response Handling¶

Manage task results - Responses Overview - Handle task outputs

💡 Best Practices¶

Task Design¶

Clear Requirements: Define specific, measurable outputs
Appropriate Timeouts: Allow sufficient processing time
Fair Payment: Match payment to task complexity

Data Structure¶

Validate Locally: Test your function and data structure first
Use Pydantic: Leverage type safety for data models
Handle Errors: Design graceful error handling

Cost Optimization¶

Batch Similar Tasks: Group related work together
Right-size Payments: Don't overpay for simple tasks
Monitor Performance: Track task success rates

🎯 Next Steps¶

Task Configuration - Detailed parameter configuration
Publishing Tasks - Complete publishing workflow
Task Templates - Ready-to-use configurations for popular sources
Responses Overview - Handle task results

Ready to run your first task! 🚀