SillyTavern with an uncensored API: setup

The settings

Open the plug icon, API Connections, and fill these in. Then press Connect.

API:                     Chat Completion
Chat Completion Source:  Custom (OpenAI-compatible)
Custom Endpoint:         https://uncensoredchats.com/v1
Custom API Key:          sk-uc-...
Model ID:                glm-5.3-flash-xploded

Your key comes from the dashboard after you sign in and add credit. Use Chat Completion, not Text Completion: these models expect a list of messages, and Text Completion sends one block of text that they answer badly.

Which model for roleplay

GLM 5.3 Flash (glm-5.3-flash-xploded) is the one to start on. Fast, cheap, a million tokens of context, so a long chat with a big character card and lorebook still fits.
DeepSeek Flash Lite (deepseek-v4.1-flash-lite-1017-xploded) is the cheapest for very long chats, because repeated history is billed at $0.03 per million tokens.
Kimi K3.1 Pro (kimi-k3.1-pro-xploded) writes the best prose. Save it for scenes that matter: output is $37.50 per million tokens.

Settings worth changing

Turn streaming on. Replies appear as they are written.
Raise the context size in SillyTavern to match the model. It defaults far below a million tokens and will trim your history for no reason.
You do not need a jailbreak prompt. Remove it. On these models it only adds tokens you pay for on every turn.

If it does not connect

401: the key is wrong or has a space pasted with it.
402: no credit, or the key reached its spend cap. Add credit in Billing.
Model not found: the id is mistyped. Copy it from the models page.
Nothing happens: check the endpoint ends in /v1 with no trailing slash.

Keep reading

Updated 12 November 2026. Prices are read from the live price list.

Start chatting All models and prices

SillyTavern on an uncensored API

The settings

Which model for roleplay

Settings worth changing

If it does not connect

Keep reading