Decentralized GPU inference on Robinhood Chain — read the docs

Demand and supply on one scheduler.

Jobs are assigned to workers as capacity frees up, verified, and settled. Idle, assigned, and verifying slots below are a live illustration of that loop.

Worker slotsScheduler sweep
Idle Assigned Verifying

Become a provider.

Run the worker client

Install the worker, point it at the orchestrator, and it benchmarks your machine automatically.

List your models

The worker reports installed models and measured throughput. The router only sends jobs it can serve.

Get paid per job

Earnings are credited per usage, not per hour. A higher bond unlocks a higher worker share.

Hold $INFP, stake for revenue

$INFP is an ERC-20 on Robinhood Chain. Holding it gates access to the network — hold $INFP to rent GPUs — and staking it earns a pro-rata share of real network revenue. Holder staking and worker bonding stay separate so neither blocks the other.

Worker bonding

Workers post collateral to access higher earning tiers. If verification fails on a job, part of the bond can be slashed.

Read the API reference