Nothing to sign in to
There is no account, so there is nothing to tie a question to. It leaves your machine already stripped of you and arrives as text and nothing else.
Open the chat →NOBODY IS LISTENING
Ask the network. A stranger's idle machine answers, and the question does not survive the reply.
There is no account, so there is nothing to tie a question to. It leaves your machine already stripped of you and arrives as text and nothing else.
Open the chat →No datacenter, no dispatcher. An idle GPU somewhere picks the job off the wire. You never learn whose, they never learn it was you.
See the network →A prompt lives exactly as long as its answer takes to stream. Nothing to leak, nothing to subpoena, nothing to train on. What you may ask is up to you.
How it works →No account, no cookie, no session worth keeping. What crosses the wire is the question and the model you chose.
Whichever idle GPU can serve your model first takes the job. There is no queue with your name on it.
The model runs on their hardware and the answer streams back a token at a time, straight into your tab.
The last token lands and the prompt stops existing, everywhere. Not archived, not logged, not sampled.
One API, no account. A single request mints a key, and the key is the account. Same models, same privacy, streamed straight back to your code.
Read the docsPoint a spare GPU at the network and it answers for strangers while you sleep. 70% of every job, paid in USDC on Arc, a dollar a point, whenever you like.
Lend a machine