A permissionless network where any device becomes a node in humanity's shared AI infrastructure — owning, training, and earning from the intelligence it helps build.
OWL is a decentralized AI training and inference network. Nodes contribute compute, earn OWL, and participate in governance — no intermediaries, no centralized control.
Lock OWL tokens proportional to your node tier. Mobile, PC, and Server nodes form a tiered mesh — each contributing what it can, earning what it deserves.
Nodes receive model shards and gradient tasks. A four-layer attestation system verifies each contribution with ZK proofs — Byzantine fault tolerant by design.
Every verified contribution earns OWL from the 600M reward pool. Emissions follow a halving schedule — early contributors earn the most.
Token holders vote on protocol upgrades via quadratic voting. No whale dominance — collective wisdom over capital concentration.
From your phone to a data center rack — the OWL network is designed to include every form of compute, each with a role proportional to its capability.
Lightweight inference tasks, edge preprocessing, and network participation from any smartphone or low-power device.
The backbone of the network. Consumer GPUs handle model sharding, gradient computation, and local fine-tuning rounds.
High-throughput orchestration nodes. Aggregates gradients, manages model state, and coordinates training across the wider mesh.
No mint function. No admin keys. Fixed 1B supply, distributed at genesis. The 600M node rewards pool halves every 365 epochs — aligning early contributors with long-term protocol health.
Inference fees generate protocol revenue, a portion of which burns OWL — creating deflationary pressure as network usage grows.
The intelligence of tomorrow is being built today, by every node that joins.