The GPU network that
pays for itself.
Simulation and skill training run on a distributed compute network. Anyone can contribute GPU capacity and earn $LOX per verified work unit; builders get sim runs at a fraction of hyperscaler prices.
Bring your GPUs
Run the provider agent on anything from a 4090 in a closet to a rack of H200s. Capacity is benchmarked, tiered and put to work within minutes.
Earn per work unit
Sim steps and training epochs are metered as verifiable work units. Payouts settle in $LOX to your address every epoch — no invoices, no net-30.
Verified execution
Work units are spot-checked by redundant execution. Providers that return wrong results lose stake and reputation; correct ones compound both.
Burst scheduling
A million-step run fans out across hundreds of providers and reassembles deterministically. Big validation sweeps finish in minutes, not nights.
Region pinning
Keep regulated datasets in-region with residency constraints — the scheduler only places your work on providers that match the policy.
Spot economics
Idle capacity is cheap capacity. Deadline-flexible jobs bid on the spot tier and routinely clear at half the standard rate.
Put your idle GPUs to work.
Install the provider agent, stake, and start earning $LOX on capacity you already own.