API
Quick start
- 1
Make a key
Generate a key - 2
Point your client here
BASE URL OpenAI-compatible
https://askuntold.ai/v1curl https://askuntold.ai/v1/chat/completions \ -H "Authorization: Bearer YOUR_API_KEY" \ -H "Content-Type: application/json" \ -d '{"model": "untold-uncensored", "messages": [{"role": "user", "content": "Hello!"}]}' - 3
Check what you were served
Every response carries an
untoldblock naming the privacy mode it was actually delivered under, and a receipt digest you can verify without us.
Rate limits
signed out 20 requests per minute on this plan. Every response carries x-ratelimit-* headers, so a client can read its real budget rather than trusting this sentence.
Use Untold from a client you already have
Anything that speaks the OpenAI chat API works: set the base URL to https://askuntold.ai/v1 and paste a key. Streaming is supported, so a client that streams by default needs nothing else.
- Cursor
- Settings → Models → OpenAI API Key → Override Base URL. Add the model id you want by hand; Cursor does not read the model list.
- Open WebUI
- Settings → Connections → OpenAI API. The model list is read from
/v1/models, which returns the models your plan can actually call — so nothing in the dropdown refuses you when you pick it. - SillyTavern
- API → Chat Completion → Custom (OpenAI-compatible), then the base URL and key.
Claude Code is not on this list. It speaks Anthropic's Messages API rather than the OpenAI one, so pointing it here does not work — a base URL is not the only difference. We would rather say that than let you find out.