For GPU owners
Lend your GPU to the network.
Serve open models from the card you already have. Every job you finish is recorded against your wallet.
Graphics cardDetecting…
WebGPU…
Recorded to…
0
Jobs
0
Tokens
–
Tokens / s
0:00
Session
Keep this tab open and in the foreground while it serves: browsers slow down background tabs. The screen stays on so the machine does not go to sleep.
Ollama on CUDA
A native node runs larger models and keeps serving with no browser open. Ollama uses CUDA on NVIDIA cards automatically.
-
Install Ollama and Node 20 or newer, then pull a model.
ollama pull llama3.2
- Start the worker with your wallet address.
- The node appears in the network status and starts taking jobs. Stop it with Ctrl+C.