Text Worker

Run LLMs on The Grid. Earn AIPG for powering chat and text generation.

What You’re Running

The Grid Inference Worker connects your local LLM backend to The Grid:

1. You run a backend (Ollama, vLLM, etc.) with your models
2. The worker connects to The Grid
3. Users chat at aipg.chat or via API
4. Your hardware generates responses
5. You earn AIPG

No Python required for pre-built binaries. Just download, run, and follow the setup wizard.

Requirements

CPU-only is possible for small models but significantly slower.

VRAM by Model Size

Quick Start

1. Install Ollama

curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3.2:3b

2. Get an API Key

Register at dashboard.aipowergrid.io

3. Download the Worker

Grab the binary for your platform from Releases:

Windows — Double-click the exe.

macOS — Unzip, open the app.

Linux — chmod +x grid-inference-worker-linux-x64 && ./grid-inference-worker-linux-x64

4. Configure via Web Wizard

The worker opens a setup wizard at http://localhost:7861:

• Enter your API key
• Select your backend (Ollama auto-detected)
• Choose a model
• Click Start

That’s it. Jobs come in automatically.

Alternative: Run from Source

git clone https://github.com/AIPowerGrid/grid-inference-worker
cd grid-inference-worker
pip install -e .
grid-inference-worker

Alternative: Docker

git clone https://github.com/AIPowerGrid/grid-inference-worker
cd grid-inference-worker
cp .env.example .env
# Edit .env with your API key
docker compose up -d

Configuration

Via Web Wizard (Recommended)

Everything configurable at http://localhost:7861.

Via CLI

grid-inference-worker \
  --api-key YOUR_KEY \
  --model llama3.2:3b \
  --backend-url http://localhost:11434

Via Environment

# .env
GRID_API_KEY=your-key
MODEL_NAME=llama3.2:3b
BACKEND_TYPE=ollama
OLLAMA_URL=http://127.0.0.1:11434
GRID_WORKER_NAME=my-worker

Install as Service

Auto-start on boot:

grid-inference-worker --install-service

Supported Backends

Anything with an OpenAI-compatible API works. Set BACKEND_TYPE=openai and point OPENAI_URL at it.

Earning

How It Works

1. User sends a message at aipg.chat or via API
2. The Grid routes it to your worker
3. Your model generates the response
4. You earn AIPG

What Earns Most

• Popular models — Llama 3, Mistral get the most traffic
• Fast responses — quicker = more jobs/hour
• 24/7 uptime — consistent availability
• Streaming — enable for real-time token delivery

Troubleshooting

Worker won’t start?

• Check Python 3.9+ if running from source
• Verify Ollama is running: ollama list

No jobs?

• Normal during low traffic
• Check worker shows online in dashboard
• Make sure you’re running a popular model

Out of memory?

• Use smaller model (3B vs 7B)
• Try quantized versions (q4_0)
• Close other GPU apps

Links