// supply compute
Earn USDF with your Mac or GPU.
Serve open models, list your own endpoint, or sell unused provider credits and API quota. Every request you serve earns USDF.
// 01 / supply paths
Three ways to supply compute.
// 02 / install
Set up a machine.
Requirements
- M1 or newer
- macOS 14+
- 8 GB memory minimum
- 32 GB+ recommended
- 50 GB free disk
- Outbound port 443 only
// 03 / models
Models you can serve.
| Model | Memory needed | Earn per 1M tokens |
|---|
Loading models…
// 04 / verification
How it's verified
01
Hidden test prompts
Every machine receives hidden test prompts, and its answers are checked against known results.
02
Proofs on GPUs
GPU machines send a proof with their work.
03
A bond at stake
Each operator posts a bond. A machine that fails its checks has the bond slashed.
// 05 / payouts
Paid in USDF.
Paid in USDF.
Every request your machine serves earns USDF at your price or the network price, sent to your wallet automatically every hour. Nothing to claim.
// 06 / questions
Questions.
What do I need?
A Mac with Apple silicon (M1 or newer) on macOS 14 or later, with 8 GB of memory at minimum and 32 GB or more recommended, plus 50 GB of free disk. Or a Linux machine with an NVIDIA GPU on driver 580 or later. The node only needs outbound internet on port 443.
Is it safe to run on my machine?
The node runs as your user, needs no admin rights and opens no ports; it only connects out. Each request arrives encrypted for your machine alone, is processed in memory and is never stored. The app blocks debuggers and checks System Integrity Protection, so requests stay private to the people who sent them.
How do I get paid?
Every request your machine serves earns USDF at your price or the network price. Earnings go to your wallet automatically every hour; there is nothing to claim.
Which models can I serve?
The table above lists every model the network serves, the memory each needs and the most it pays. The app offers the ones that fit your machine and downloads them for you.
How do I remove it?
Quit it from the menu bar or the command line, then run its uninstall command (shown in the app's help). It removes the app, its background service and every downloaded model.