PRIVATE AI, RUN BY NO ONE

an0n

Type a question and a stranger's idle GPU answers it. The network forgets you the moment it replies.

>
01 WHAT YOU GET
SENDER · UNKNOWN

Ask like no one's reading

Because no one is. There's no signup and no session to tie a question to — it leaves your machine already stripped of you.

ROUTED WORLDWIDE

Answers from nowhere in particular

No datacenter to knock on. Idle machines around the planet pick jobs off the wire — you never learn whose GPU, they never learn it was you.

WIPED ON DELIVERY

Deleted by default

A prompt lives exactly as long as its answer takes to stream. Nothing to leak, nothing to subpoena, nothing to train on.

NO GATEKEEPER

No one to tell you no

Open models on independent hardware — no corporate layer deciding which questions are acceptable. What you may ask is up to you.

02 THE LIFE OF A PROMPT
  1. t + 0.0s

    It leaves clean

    Scrubbed of anything that could trace back before it touches the wire — no account, no session, no name.

  2. t + 0.3s

    The network decides

    No dispatcher, no datacenter — the job falls to whichever idle machine can serve your model first.

  3. t + 1.2s

    A stranger thinks

    Someone else's GPU runs the model and streams the answer back to you, a token at a time.

  4. t + done

    Then nothing

    The instant the last token lands the job is gone everywhere. No transcript outlives the request.

03 TWO WAYS TO PLUG IN
FOR DEVELOPERS

Build on the network

One API, no account, no data center. Mint a key with a single request and route prompts to a live mesh of GPUs — same models, same privacy, streamed straight back.

Read the docs
FOR GPU OWNERS

Plug in, get paid

Point a spare GPU at the network and it answers for strangers while you're away — 70% of every job, paid in $USDG, a dollar apiece and cashable to $ANON whenever you like.

Start earning
machines on the wire
open models serving
0prompts ever stored