---
title: "Xoul AI proxy: custom engine setup, uncensored LLMs"
canonical: https://uncensoredllm.ai/guides/xoul-ai-proxy
generated: 2026-10-27T06:25:24.937Z
source: https://uncensoredllm.ai/guides/xoul-ai-proxy
base_url: https://api.uncensoredllm.ai/v1
---

# Xoul AI custom engine (proxy) with uncensored models

Xoul AI proxy setup on the Generic provider with an uncensored roleplay build, inside Xoul's 875-token and 3,500-character reply limits.

Steps checked October 24, 2026 · prices as of October 27, 2026

A Xoul AI proxy is a custom engine: the Generic provider, a base URL, a key and a model name. This guide fills those fields with an uncensored roleplay build and sets the engine inside Xoul's own limits, 875 tokens and 3,500 characters per reply. The field names come from Xoul's official documentation, read on 24 October 2026.

If builds with the refusals edited out are new to you, start with [AI with no restrictions](https://uncensoredllm.ai/blog/ai-with-no-restrictions), then come back here to connect one.

## What Xoul AI is

Xoul AI is a site for creating and chatting with AI characters, immersive scenarios and original worlds. Out of the box a chat runs on the models Xoul provides.

A custom engine is Xoul's name for a proxy. It sends the chat to a model on an API you choose, with your own key. Xoul announced the feature on its [official documentation](https://xoul-ai-official-documentation.gitbook.io/xoul.ai-official-guide): you add a custom engine profile from your conversation, under the model selector. The fields are described on the [Custom Engine (Proxy) page](https://xoul-ai-official-documentation.gitbook.io/xoul.ai-official-guide/navigation-and-information/navigation-and-interfaces/chat-interface/custom-engine-proxy). Both were read on 24 October 2026.

## The Xoul AI proxy fields, filled in

These are the values for this API. The labels are the ones in Xoul's Custom Engine (Proxy) doc.

- Engine Name: UncensoredLLM Any name; Xoul saves the engine for reuse
- Provider: Generic Shows the Base URL field
- Base URL: https://api.uncensoredllm.ai/v1 The base URL, ending in /v1
- API Key: sk-ullm-live-… Created in the console, shown once
- Model Name: cydonia-24b-v4.3 Any id from the Models page
- Max Tokens: 875 Xoul's documented ceiling

Xoul's doc does not say what form the Generic Base URL takes. The base URL ending in `/v1` is what OpenAI-compatible clients expect, and it is the form this API documents for every client on the [docs page](https://uncensoredllm.ai/docs).

## Connect Xoul AI to the uncensored API

1. Create a key in the [console](https://app.uncensoredllm.ai/console). It starts with `sk-ullm-` and is shown once, so paste it somewhere safe first. The balance is prepaid from $10; a key with nothing on it is accepted but the first message answers 402.
2. Pick a model id from the [uncensored model list](https://uncensoredllm.ai/models) or from the table below. Copy it rather than typing it.
3. Open a conversation in Xoul. Under the model selector, add a **custom engine** profile. This path is from Xoul's announcement; the Custom Engine (Proxy) page shows the form but not the click path.
4. Type a name in **Engine Name**, for example UncensoredLLM. Xoul saves the engine under that name so you can reuse it without filling the form again.
5. Set **Provider** to **Generic**. The **Base URL** field appears. Paste `https://api.uncensoredllm.ai/v1`.
6. Paste the key into **API Key**. Xoul's doc warns not to share it: anyone holding it can spend your balance.
7. In **Model Name**, select the id from the list, or type it and select it, for example `cydonia-24b-v4.3`.
8. Optional: write a **System Prompt**, and set **Max Context Length** from the table in the next section.
9. Set **Max Tokens** to 875 or lower. Leave the other sampling fields (Temperature, Top P, Seed, Top K, Min P, Repetition Penalty) at their defaults for the first message.
10. Save the engine and send a message. A reply in the chat means the URL, the key and the model id are all right.

## Limits Xoul sets on a custom engine

Three limits come from Xoul's own doc, not from the API. Plan the engine around them.

- **Max Tokens should not exceed 875.** That is Xoul's recommendation for the reply length. A model told it may write 875 tokens stops at 875, mid-sentence if needed.
- **A reply is limited to 3,500 characters.** The doc says a proxy can generate more than that, but a reply longer than 3,500 characters cannot be saved after you edit it. The two limits line up at about four characters per token, a common rough ratio for English text.
- **Response Styles and creator system prompts are off.** In the doc's words, proxies can leak that information, so Xoul does not offer them with a proxy. Put the instructions you need in the engine's own System Prompt field.

## Which uncensored build to pick for Xoul

All three builds below carry the roleplay tag in the catalog and take the OpenAI wire format. The prices and windows are read from the live catalog when the page renders.

| Model | Build | Input $/M | Output $/M | Context | Wire formats |
| --- | --- | --- | --- | --- | --- |
| [Cydonia 24B v4.3](https://uncensoredllm.ai/models/cydonia-24b-v4.3) cydonia-24b-v4.3 | Roleplay | $0.484 | $0.60 | 131K | openai |
| [Qwen3.8 27B Antislop](https://uncensoredllm.ai/models/qwen3.8-27b-antislop) qwen3.8-27b-antislop | Roleplay | $0.35 | $2.10 | 262K | openai |

Prices per million tokens, read from the live catalog on 2026-10-27.

| Model id | Context window in the catalog | Max Context Length to enter |
| --- | --- | --- |
| [`cydonia-24b-v4.3`](https://uncensoredllm.ai/models/cydonia-24b-v4.3) | 131,072 tokens | 32768 |
| [`qwen3.8-27b-antislop`](https://uncensoredllm.ai/models/qwen3.8-27b-antislop) | 262,144 tokens | 32768 |

Context windows read from the live catalog on 2026-10-27. The value to enter is the window capped at 32,768, or 16,384 where no window is published.

Never enter more than the window. Xoul's doc also notes that a smaller context limit lowers the cost of each call, since the whole context is sent as input on every message. `qwen3.8-27b-antislop` is one of the Qwen 3.8 tunes; the rest of that family is on the [uncensored Qwen page](https://uncensoredllm.ai/uncensored/qwen).

## What an evening on Xoul costs

The worked example: 200 messages in one sitting. Every reply is priced at the full 875-token ceiling, so this is the most the evening can cost at each context size. Real replies are usually shorter. The short column sends about 4,000 tokens of character and history per message; the long column sends 16,000.

| Model | 200 messages at 4,000 context | 200 messages at 16,000 context |
| --- | --- | --- |
| [Cydonia 24B v4.3](https://uncensoredllm.ai/models/cydonia-24b-v4.3) | $0.49 | $1.65 |
| [Qwen3.8 27B Antislop](https://uncensoredllm.ai/models/qwen3.8-27b-antislop) | $0.65 | $1.49 |

Computed at render from the catalog prices on 2026-10-27: messages × prompt tokens × input price, plus messages × 875 × output price.

Input still dominates: 200 × 4,000 is 800,000 input tokens against at most 175,000 output tokens. A lower Max Context Length moves the bill more than the choice of model. Every per-token rate is on the [pricing page](https://uncensoredllm.ai/pricing).

## Check the key and the model id before Xoul

Before you fill in the engine, this request lists the model ids your key can reach. It costs nothing and takes a second. Swap in your key and run it from a terminal.

```bash
curl https://api.uncensoredllm.ai/v1/models \
  -H "Authorization: Bearer sk-ullm-live-…"
```

A 200 with a list of ids means the key is good. If the id you chose is in that list, the only field left to get wrong is the Base URL. A 401 means the key, and the message says which way:

```
HTTP 401
{"error":{"message":"No API key was sent.","type":"authentication_error","param":null,"code":"invalid_api_key"}}
```

```
HTTP 401
{"error":{"message":"The API key is not valid.","type":"authentication_error","param":null,"code":"invalid_api_key"}}
```

Both bodies as the API returned them on 24 October 2026.

## Troubleshooting Xoul custom engine errors

### 401: the key

"No API key was sent." means the API Key field is empty. "The API key is not valid." means it is wrong, cut short or revoked. Create a fresh key in the console and paste it whole.

### 404 model_not_found: the model id

The Model Name is not in the catalog: a typo, a display name instead of an id, or a build that has since left. Run the request above and copy the id from its output.

### 402: the balance

The prepaid balance is at zero. Top up in the console; a request refused with 402 is not charged.

### Browser errors (CORS)

The API allows calls from any origin, including the Authorization header (checked on 24 October 2026). Whether a request comes from your browser or from Xoul, there is nothing to switch on.

### A reply cut off mid-sentence

Not an error. The reply reached Max Tokens. Raise it, up to 875, or ask for shorter replies in the System Prompt.

The [daily real-model check](https://uncensoredllm.ai/proof) sends a fixed prompt to every model each day, so the model behind an id is the one on its label.

## Questions

### Is Xoul AI free?

Xoul’s own plans are listed on xoul.ai. A custom engine bills the model use to the API you connect: here per token, from a prepaid balance that starts at $10. See per-token prices.

### What does an AI proxy do?

It sends your chat to a model on another API instead of the app’s own models. On Xoul that is a custom engine: a base URL, a key and a model name. The app keeps your characters and chats; the API runs the model.

### Do I want proxy on or off?

On, when you want a model the app does not offer, such as an uncensored roleplay build, billed to your own key. Off, to use Xoul’s built-in models, Response Styles and creator system prompts, which Xoul turns off while a proxy is in use.

### What Max Tokens should I use on Xoul?

No more than 875, as Xoul’s doc recommends. Xoul limits a reply to 3,500 characters, and a longer reply cannot be saved after an edit.

### Which uncensored model works best on Xoul?

Any id that takes the OpenAI wire format works with the Generic provider. For character chat, start with a build the catalog tags roleplay; the table above lists three with their windows and prices. See the uncensored model list.

### Is this different from a jailbreak prompt?

Yes. The builds here are abliterated or uncensored fine-tunes, so the change is in the weights, not in a prompt, and there is no jailbreak text to paste into the System Prompt. See AI with no restrictions, explained.

[Get a key](https://app.uncensoredllm.ai/console) Prepaid from $10, no subscription.

## New models and price changes, once a month.

New uncensored builds with their prices, price moves, and what the daily model check found.