Uncensored AI for roleplay
Fiction needs villains, violence, grief and desire. Uncensored AI writes the scene you asked for instead of stepping out of it to refuse.
Why mainstream models break character
Assistant-tuned models were trained to back away from exactly what fiction is made of, which is why stories on them drift towards safe, soft endings and characters suddenly talk like a help desk. An uncensored model keeps the base model's writing ability and drops the urge to steer.
Connect SillyTavern
- In SillyTavern open API Connections and set API to Chat Completion.
- Set Chat Completion Source to Custom (OpenAI-compatible).
- Custom Endpoint:
https://wildwestapi.com/v1, and your key in Custom API Key. - Connect, pick a model, and send a message.
The SillyTavern setup guide covers every box and the settings that hold up over a long story. Sites that call an API straight from your browser tab do not work, because the API does not answer cross-origin requests; run the chat through SillyTavern instead.
Which model to start with
| Model | Kind | Context | Input / 1M | Output / 1M | Tools | Images |
|---|---|---|---|---|---|---|
| Outlaw 1 Xploded outlaw-1-xploded | Abliterated | 1M | $0.30 | $1.00 | Yes | Yes |
| GLM 5.3 Xploded glm-5.3-xploded | Uncensored | 1M | $2.50 | $4.50 | Yes | No |
| GLM 5.3 Flash Xploded glm-5.3-flash-xploded | Uncensored | 1M | $0.40 | $1.60 | Yes | Yes |
| MiMo V2.6 Flash Xploded mimo-v2.6-flash-xploded | Abliterated | 1M | $1.00 | $3.00 | Yes | Yes |
| Qwen3.8 27B Xploded qwen3.8-27b-xploded | Uncensored | 512K | $0.30 | $2.40 | Yes | Yes |
Every model on the line holds at least 512K tokens of context and up to 1M, so a long campaign keeps its history. Outlaw 1 Xploded is the cheapest per output token. The roleplay model guide goes deeper into choosing.
What a long chat costs
The whole history goes back to the model with every message, so input is most of the bill. A scene of 6,000 tokens of card and history with a 300 token reply, on a model at $0.40 per million input tokens and $1.60 per million output, costs about $0.003 a message. Capping the context size in your app keeps it flat as the story grows.
Moving from a hosted character site? Chub AI and Crushon AI have their own guides.