Decentralized GPU inference on Robinhood Chain — read the docs
Kunagi Systems

Core concepts

The five terms that show up everywhere else in these docs, and which build phase each belongs to.

ConceptMeaningLaunch status
Inference requestA prompt and model selection sent by a user through the app or API.Phase 1
OrchestratorThe routing service that queues jobs, selects workers, streams responses, and records usage.Phase 1
WorkerA browser or native client that runs inference on a GPU and returns streamed tokens.Phase 2
Bootstrap supplyOwned or rented GPUs used so the product works before marketplace supply is deep.Phase 1
Worker bondCollateral posted by workers when verification and slashing are introduced.Phase 3
PreviousQuick start