Overview

Kunagi Systems is a decentralized inference network. Users send inference requests, Kunagi routes each request to available GPU supply, and workers earn for the compute they contribute.

Decentralized inferenceGPU marketplaceOpen modelsPrompt retention: none by default

The network is built around open models, short lived prompt handling, and a marketplace for independent GPU providers. Kunagi is described publicly as a network and marketplace. Internally, launch supply can include rented or owned GPUs while the worker marketplace is being built.

Worker mesh

Each point in the mesh below represents a worker or bootstrap GPU. Links represent eligible routing paths by model, capacity, and observed latency.

Worker registry · live