Private AI
AI that doesn't take notes.
Pick a model and ask. Venice's private models keep zero data by contract, and our gateway doesn't store your prompt or the answer. Once the answer reaches you, that's it.
Your daily AI, left today
—
Answers left today
The model's answer, streamed as it's written, with what it cost.
Ready when you are. Ask anything, or pick a prompt.
The exact call your app or agent would send for this answer. Swap in your own rtn_ key and it runs.
POST /v1/chat/completions
What comes back with the answer
The gateway holds the request plus max_tokens at the output price, then charges the real token count. Daily AI pays cost. Paid credit pays 1.3×.
Pay as you go
When today's AI runs out, keep going on credit. Pay in USDC or USDG: one dollar buys a dollar of credit, and calls on credit cost 1.3× Venice's price. Part of every payment burns $NOWHERE.
Hand it to your agent
Issue a key with the ration scope and give it to your agent. It can only spend your daily AI, never your paid credit. Revoke it whenever you like.
curl Your agent asks
curl https://api.nowhere.example/v1/chat/completions \
-H "Authorization: Bearer rtn_••••••••••••" \
-H "Content-Type: application/json" \
-d '{"messages":[
{"role":"user","content":"Plan my day"}]}'It speaks the OpenAI format. Leave out model and the gateway picks the cheapest one. When today's AI runs out, the call is refused. It never touches your paid credit.
What happens to your prompt
- Your deviceSends it to our gateway over TLS.
- Our gateway, on CloudflareChecks your allowance and passes the call on. It writes down your wallet, the model, the token counts and the cost. It doesn't store the prompt or the answer.
- Venice's private modelZero data retention, by contract. Web search and Venice's own system prompt are switched off.
- Back to youThe answer streams to your screen. Your allowance pays Venice's price.
- Then nowhereWe keep no copy, and Venice's private models keep none under its terms.
Venice, in its own words
“Private models have zero data retention.”
“Zero data retention, contract-enforced.”
From docs.venice.ai, read in October 2026. We only list Venice's private, open-weight models.
Prices are Venice's, from our gateway's catalog. The private AI opens soon. Until then, the preset prompts show example answers.