Launch
Run a GPU

Plug in a GPU, earn $ORBIT

One command connects Ollama, LM Studio, vLLM or llama.cpp to ORBIT, and one click in your browser approves it. Serve public missions and earn $ORBIT for every accepted call, or run a private rig that serves only you, for free.

See every station
Stations online00
Paid to operators0.00
Probe pass rate--
Connect this machine

Earnings accrue per accepted call and are paid to your wallet in verified batches.

1
Insert coin
2
Public or private?
3
Run one command on the machine with the GPU
curl -fsSL https://api.runorbit.xyz/install.sh | sh

It finds Ollama, LM Studio, vLLM or llama.cpp on this machine, shows a pairing code and opens this site to approve it, then keeps running in the background. Python 3.9+ only; outbound connections only, so it works behind a home router. No model server yet? Install Ollama and run ollama pull qwen3:8b.

The link the worker prints does this for you.
Waiting for your first rig
Advanced · register with a token instead of pairing
Models in demand

What missions and chats asked community GPUs for in the last 24 hours. "Unmet" calls fell back to hosted APIs because no station could serve them, so serving those models earns the most.

Model · tierPays24hUnmetGPUs
No community demand recorded yet.
Where your $ORBIT goes
FuelFunders load missions with credits (1 $ORBIT = 100 credits).
90%Every community-served call pays its operator. Bigger model tiers pay more per token.
10%Protocol fee. 2% of it goes to a mission's original creator when forks run.
FallbackHosted-API calls pay the API bill when no station can serve.
ProbesOperators get unannounced known-answer checks, harder for bigger model tiers. Wrong answers cost reputation.
Own rigYour own rigs serve you first at zero cost, and earn nothing from you (there's nothing to farm).