Our miner works in the gaps.
The Duplex Miner is Peryn’s own Pearl miner. On a GPU that serves AI, it mines between requests. Available with our datacenter runtime.
Mining while it serves.
Mining during serving is switched on only where we measured it adding under 5% to response times (under 5 ms at low load), so your customers are served as before. Available with our datacenter runtime.
How it decides, GPU by GPU: Orchestration. Mining with your own rigs today: Mine with Peryn.
- Serving first
- It pauses while any request is unfinished and never runs inside one.
- Two incomes
- USD for every token your GPU serves, PRL for every share it finds.
- Inside your scheduler
- Kubernetes, Slurm or bare metal, with your rules for when and how much.
- Measured per GPU
- Measured on H100 and H200. Other cards are added once they pass the same bar.