SillyTavern API, uncensored.

An uncensored DeepSeek build and a roleplay tune in SillyTavern, on one key. Base URL, key, model id: two minutes.

Connect SillyTavern

  1. Make a key in the console and top up from $10. The key starts with sk-unr-.
  2. In SillyTavern, open API Connections (the plug icon).
  3. API: Chat Completion. Chat Completion Source: Custom (OpenAI-compatible).
  4. Custom Endpoint (Base URL): https://api.unrestrictable.ai/v1.
  5. Custom API Key: your key.
  6. Model ID: deepseek-v4.1-flash-1017-uncensored or artemis-v1.1.
  7. Press Connect and send a test line. Leave streaming on.

DeepSeek API in SillyTavern, uncensored

SillyTavern has its own DeepSeek source, which talks to the stock model and its refusals. deepseek-v4.1-flash-1017-uncensored is a DeepSeek build with the refusals tuned out, reached through the Custom source above: smart and fast. More builds of the same family are on the DeepSeek page.

A roleplay tune for long chats

artemis-v1.1 is a fiction-tuned model that holds a character over long chats, with a long context. Switch between the two ids in the Model ID field with no other change. Set SillyTavern’s Context Size no higher than the model’s window on its model page.

Instead of a free OpenRouter API

Free routes are a fine way to start, but they are shared and rate limited. Here you pay per token from a $10 top-up with no plan, every model is an uncensored or roleplay build, and a request that fails costs nothing. Try any of them first in the browser chat, with three free turns and no account.

The two models, priced live

Uncensored models and their prices per million tokens
Wire formats
Uncensored DeepSeek V4.1 Flash 1017deepseek-v4.1-flash-1017-uncensoredHouse build$0.8003$3.2011049Kopenai
Artemis v1.1artemis-v1.1Roleplay$0.35$1.575262Kopenai

Prices per million tokens, read from the live catalog on 2026-11-21.

The full walkthrough, with sampler settings, eight models and an importable preset, is in the SillyTavern guide.

FAQs

Questions.

https://api.unrestrictable.ai/v1, under Chat Completion with the Custom (OpenAI-compatible) source. Keep the /v1; SillyTavern adds the rest.

Yes. Pick the Custom (OpenAI-compatible) source and type deepseek-v4.1-flash-1017-uncensored as the model id. It is an uncensored DeepSeek build, so it answers where the stock model declines.

The API runs on a prepaid balance from $10, priced per token, with no plan. To try the models first, the browser chat gives three free turns with no account, and a free account gets 10 messages a day.

Free routes elsewhere are shared and rate limited. Here every one of the 369 chat models is an uncensored or roleplay build, on one key, and a request that fails is not charged.

artemis-v1.1 holds a character over long chats. deepseek-v4.1-flash-1017-uncensored is smart and fast, a little plainer in its prose.

Try the models before you connect them.

Three free turns in the browser, no account.

Join on TelegramHelp