How it works
Three steps: miners contribute compute, the pool serves inference, and miners earn credits for it.
1
Miners contribute GPU
GPU owners join the pool and run inference workloads. Any supported GPU — from data center H100s to consumer cards.
2
Pool serves AI inference
The pool routes API requests to available miners. OpenAI-compatible endpoints — just change the base URL and it works.
3
Miners earn, customers save
Miners get 75% of revenue in credits. Customers get OpenAI-compatible inference with zero data retention.