The measured one
Private
A raw hop still reconstructs the prompt. One MPC share is noise. The bound costs 2.42 MiB per token. The figure is in the Paper.
Meet me behind
the old library
at midnight.
recover 100%
0.21% · chance
Private inference network
The model is public.The prompt is not.
Uncensored and decentralized - the first network built to prove an untrusted node cannot recover what you typed.
Share lattice · Game of Life
01 / The problem
Sharding solved size. Not privacy.
Split a model across strangers' GPUs and every node still touches raw activations - enough to reconstruct pieces of a prompt.
02 / How it leaks
A request crosses machines you don't control.
Every node handles real activation tensors. Cloak exists to quantify that exposure - and close it.
Decentralized
Anyone's GPU, pooled into swarms. No datacenter.
Uncensored
Any model. No content-policy layer in the way.
Privacy isn't a feature to bolt on later. It's the open problem - and the one Cloak exists to close.
03 / Public vs private
Public models. Private prompts.
private run
one share is noise
0.21% · chance
APIhosted100%logged
WEIGHTSlocal100%raw
OBFUSCopen100%invertible
SHARDmulti94-100%activations
CLOAK2PC0.21%chance
04 / Product
Three ways into the network
Recommended
Native Worker
Background GPU via Ollama or vLLM. The biggest models, the highest pay.
Top rate / token
npx @cloakaixyz/worker --token YOUR_TOKENZero install
Browser Worker
In-tab inference via WebGPU. One click, no terminal.
Entry rate / token
Open browser workerUser side
OpenAI-compatible API
Drop-in chat completions. No logging, pay per token.
Per token
curl https://api.cloakai.xyz/v1/chat/completionsPrivate inference economy
The $CLOAK token
What crosses the chain is metering. What never leaves the job is your prompt.
Metered, not monitored
Tokens are the billing unit. No prompt history, no profiles.
Treasury funds privacy
30% of every settlement underwrites the research that proves it.
We give a FAQ
cloak ai