Yaarbal docs
Yaarbal is a network where anyone can rent out an idle GPU by running a lightweight app, earn credits while the machine serves LLM inference for others, and spend those credits on streaming inference from the web app or the command line, with one identity on every surface. A central coordinator tracks credits and peers while inference itself runs on the GPU owner's machine. Yaarbal is in a closed, invite-gated beta.
Guides
- CLI Quickstart — Install the CLI, run a seeder, spend from the terminal.
- Agentic Proxy — Point Cline and other coding agents at the network.
- OpenAI SDK — Use the OpenAI client against the capsule.
- Personal Cloud — Your own GPU, reachable from a second device.