Leaving SillyTavern? Uncensored AI models instead
SillyTavern is a front end, not a model, so two different questions hide behind this search: a different front end, or a better backend for the one you already have. The front ends people move to are compared below, and if it is the replies you want to change, Wild West API is an OpenAI-compatible backend SillyTavern connects to in five clicks.
What SillyTavern is
SillyTavern carries no model of its own, so every refusal a user sees comes from whichever backend it is pointed at. It accepts any OpenAI-compatible endpoint.
Other options people compare with SillyTavern
| Option | What it is |
|---|---|
| RisuAI | Open-source front end that runs in the browser or as a desktop app, with character cards, lorebooks and its own scripting. Bring your own API key. |
| Agnaistic (Agnai) | Open-source roleplay front end, used at agnai.chat or self-hosted, with multi-character chats and memory books. Connects to the backend you configure. |
| KoboldCpp | Single-file local runner built on llama.cpp, with the KoboldAI Lite interface and an OpenAI-compatible API built in. |
| Janitor AI | Hosted character platform with a large public card library. Has its own free model and also takes an OpenAI-compatible proxy key. |
| Chub AI | chub.ai: a large public library of character cards, with its own chat front end and paid inference plans. Cards download in the standard format other front ends import. |
| Backyard AI | Desktop app that runs characters against local models on your own machine. Nothing leaves the device; the model has to fit it. |
What Wild West API carries instead
Wild West API sells 5 uncensored models on one OpenAI-compatible endpoint, every one able to call tools. Ordered for this particular job rather than by a single ranking: MiMo V2.6 Flash Xploded leads because it carries the longest window of the set at 1.05M, which is what keeps a long session coherent.
| Model | Context | In | Out | Tools |
|---|---|---|---|---|
| MiMo V2.6 Flash Xploded mimo-v2.6-flash-xploded |
1.05M | $1.00 | $3.00 | Yes |
| Outlaw 1 Xploded outlaw-1-xploded |
1.05M | $0.300 | $1.00 | Yes |
| GLM 5.3 Flash Xploded glm-5.3-flash-xploded |
1.05M | $0.400 | $1.60 | Yes |
| GLM 5.3 Xploded glm-5.3-xploded |
1.05M | $2.50 | $4.50 | Yes |
Using SillyTavern with Wild West API
SillyTavern takes an OpenAI-compatible key, so you can keep it and change only the model behind it.
- Open API Connections (the plug icon in the top bar).
- Set API to Chat Completion.
- Set Chat Completion Source to Custom (OpenAI-compatible).
- Paste the base URL into Custom Endpoint (Base URL) and your key into Custom API Key.
- Enter the model id, then press Connect.
| Base URL | https://wildwestapi.com/v1 |
|---|---|
| Full endpoint, if it asks for one | https://wildwestapi.com/v1/chat/completions |
| API key | Your Wild West API key, sk-ww-... |
| Model | An id from the table above, for example outlaw-1-xploded |
Switching takes one base URL
Anything that speaks the OpenAI chat completions format works unchanged. Point it here, use a Wild West API key, and name the model from the table.
curl https://wildwestapi.com/v1/chat/completions \ -H "Authorization: Bearer sk-ww-..." \ -H "Content-Type: application/json" \ -d '{ "model": "mimo-v2.6-flash-xploded", "messages": [{"role": "user", "content": "Hello"}] }'
FAQ
Is Wild West API actually uncensored, or does it just say so?
Wild West API does not moderate requests. It prices the call, holds the money against your key cap and passes the request to the model. What comes back is whatever that model does: an uncensored build refuses far less than a frontier model but can still decline, and only an abliterated build has the refusal direction removed outright. Neither tag is a promise that any given answer will arrive.
What does it cost compared with SillyTavern?
You pay per token at the rates in the table above, with no subscription. $1.00 in and $3.00 out per million tokens for MiMo V2.6 Flash Xploded.
Can I cap what a key is allowed to spend?
Yes, and the cap is enforced before the request goes upstream rather than reconciled afterwards. The reply cannot physically cost more than the hold placed before it was sent.
Will it work with my existing client?
If the client takes an OpenAI-compatible base URL, yes: point it at https://wildwestapi.com/v1 with a Wild West API key. Claude Code and the Anthropic SDKs use the Anthropic format, which is served too.