You shouldn't talk too openly to your current AI. Here, you can.

Ask anything.
Nobody's reading.

23 models including GPT, Claude, Grok and uncensored open models. No account, no phone number, and on Pro nothing you type is ever stored. 25 free questions a day, right now.

Live · 25 free questions left today
Ask something other AIs refuse, or something you'd rather not have on record.

25 free questions every day, no sign-up, on the fast open models. Pro unlocks all 23 models, removes the limit and switches on zero-retention mode.

No account, no email, no phone23 models incl. GPT, Claude, Grok, DeepSeekPro: nothing stored, ever14-day money-backCancel in one click
Models from
OpenAIAnthropicxAIDeepSeekZhipu AI
questions answered
23models
0chats stored on Pro
25free / day, no sign-up

Two rules: it doesn't spy on you, and it doesn't censor the AI

Everything else follows from those two. Big-name assistants are built to keep, review and learn from what you type; this is built to do the opposite.

Uncensored by default

Open-weight models without the corporate refusal layer. Medical, legal, financial, adult, political: you get an answer, not a lecture.

Zero data retention on Pro

Pro conversations are never written to a database. There is nothing to leak, subpoena, sell or train on.

Never trained on your data

Your prompts don't improve anyone's model. Not ours, not the model providers'. Identifying metadata is stripped before any request leaves our relay.

No account. No phone. No email.

Twenty-five questions a day before you tell us anything about yourself. Pro checkout needs an email for the receipt, and that's it.

Frontier and open models, one place

Switch between uncensored open-weight models and frontier models like Claude in a click. New models are added as they ship.

Conversations live on your device

Chat history stays in your browser, not on our servers. Close the tab and it's gone; keep it open and it's yours alone.

Encrypted in transit

TLS from your device to our relay and from our relay to the model. No plaintext on the wire, anywhere.

Long context, long answers Pro

Extended context windows for deep work and long-running conversations. Full-length answers, no truncation.

Code and agents workspace Coming to Pro

Write and run code, work on whole projects and build agents inside the same private workspace, with the same zero-retention rules. Shipping to Pro members in stages.

Cut out the middleman

Big Tech AI is a costly intermediary that sees everything you type. We run a thin relay to open models; only you see the request and the response.

Built on infrastructure that already serves 35 million requests a day.

Unrestricted runs on the same relay and payment stack as our other products. The traffic, the uptime and the abuse handling are already solved; this is a new front door, not a new back end.

35M+requests per day across the group
23models, uncensored open-weight and frontier
0conversations stored on Pro
14 daysmoney-back on every charge

Use it where you already talk

Live on the web today. Pro members get Unrestricted inside their messaging apps as each channel ships.

WebLive now
TelegramComing to Pro
WhatsAppComing to Pro
iMessage / SMSComing to Pro
APIComing to Pro

Unrestricted vs your current AI

What happens to your words after you hit send.

UnrestrictedTypical AI assistant
Stores your conversationsNever (Pro)Yes, by default
Uses your chats to train modelsNeverOften, unless you opt out
Human review of your chatsImpossible, nothing is keptYes, for "safety"
Refuses "sensitive" questionsNoFrequently
Requires account, phone or email to tryNoYes
Model choiceOpen-weight + frontierOne vendor
Can be subpoenaed for your historyNothing to hand overYes

Why this exists. Every mainstream AI assistant is a middleman. It sits between you and the model, reads everything, keeps most of it, and decides which of your questions deserve an answer.

We think an AI you can't speak to openly is a broken tool. So we removed the parts that make it unsafe to be honest: the account, the logs, the training pipeline, the refusal layer. What's left is a model, a relay, and you.

Questions

What exactly do you store?

Pro conversations: nothing, ever. Free demo questions: the question text only, with no IP address, cookie or identity attached, kept for up to 30 days so we can see what people ask and improve the product. Chat history you see on screen lives in your browser and is never uploaded.

What does "uncensored" mean here?

The default models are open-weight models without the extra refusal layers big vendors add. They won't help with anything illegal, but they won't refuse medical, legal, financial, adult or political questions either.

Which models can I use?

23 models today: uncensored open-weight models (GLM 5.3, DeepSeek V4) plus frontier models from Anthropic (Claude Sonnet 5, Opus 5, Fable 5.1), OpenAI (GPT-6 Astra, GPT-5.6, GPT-5.5) and xAI (Grok 4.6). Pro unlocks all of them without limits.

Do I need an account?

No. Ask 25 questions a day with nothing but a browser. Pro checkout asks for an email so we can send a receipt and let you cancel.

Will it work in WhatsApp or Telegram?

Web is live today. Telegram, WhatsApp and iMessage access ship to Pro members as each channel is ready; you'll be notified at the email on your receipt.

Who handles payment?

Cards, Apple Pay, Google Pay and PayPal go through Paddle, a merchant of record used by thousands of software companies. Crypto (USDT, BTC, ETH, TON, SOL, XMR and more) goes through 2328. We never see your card number or wallet keys. Cancel anytime from your receipt email; 14-day money-back on every charge.

Ask what you'd never Google.

25 questions a day, free. No sign-up.

Ask your first question