> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aurinfer.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Engine

> Configure the AI models that power your character's responses — primary engine, fallback engine, voice processing model, and advanced settings.

# AI Engine

This is where you choose **which AI brain** your character uses. Different AI models have different personalities, strengths, and speeds. You can also control how the AI writes (through Advanced Settings).

**How to get here:**

> Go to your character's dashboard → left sidebar → click **Universal** → click **AI Engine**

The full path is: **Universal → AI Engine**

This section has **three sub-pages** (shown as tabs or links inside the AI Engine page):

1. **AI Engines** — choose the main AI model
2. **Voice Processing** — choose the AI model used during voice calls
3. **Advanced Settings** — write global instructions and fine-tune how the AI generates text

***

## AI Engines

**Full path: Universal → AI Engine → AI Engines**

This is the first page you see when you open AI Engine. This is where you pick which AI model (the "brain") powers your character's text responses.

### Engine Selection

| Setting                        | What it does                                                                                                                                                                                                                          |
| ------------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Selected Engine** (dropdown) | This is the **main AI model** your character uses to write all its replies. Click the dropdown to see all available models and switch to a different one.                                                                             |
| **Fallback Engine** (dropdown) | This is a **backup AI model**. If the main model is down or broken, the character automatically switches to this one so it keeps responding. Think of it like a backup generator. Defaults to a reliable model like Gemini 2.5 Flash. |

Next to each engine's name you may see small icons that show what it can do — for example, whether it can see images (vision support) or handle very long conversations (extended context).

### Send Multiple Messages

**Send Multiple Messages** is a toggle (on/off switch) on this page.

* **When ON:** Your character can split one reply into multiple short messages sent one after the other — like how a real person sends a quick message, then follows up with another. Makes conversations feel much more natural.
* **When OFF:** The character always sends one single block of text.

### Requesting a Custom Model

If you want a model that isn't listed in the dropdown, scroll to the bottom of the page and click **Request Model** to submit a request to the AurInfer team.

<Info>
  Some of the most powerful models (like GPT-5.6 and other Top-tier models) are only available on BrahMos plan and above. If you see a model grayed out with an **Upgrade** button next to it, you need to upgrade your plan to unlock it. You can do that from **Account → Billing**.
</Info>

***

## Voice Processing

**Full path: Universal → AI Engine → Voice Processing**

This page controls which AI model is used **specifically when your character is in a voice call** (like in a Discord voice channel).

**Important:** This is different from the voice engine!

* The **Voice Processing model** decides *what to say* during voice calls (the words/text).
* The **Voice Engine** (under a different section) decides *how it sounds* (the actual voice and audio quality).

| Setting                      | What it does                                                                                                                                                         |
| ---------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Current Model** (dropdown) | The AI model that writes your character's responses during voice calls. Switch this to a faster or smarter model depending on what you need. Example: Llama 3.1 70B. |

On the right side of this page, you'll also see a **Fastest Performing Models** panel — this shows you which models are currently responding fastest based on real-time data. Use this if voice calls feel slow or laggy.

<Info>
  **Override priority:** If you have set engine rules in your **Server's Dedicated Engines** settings or in your **Account → Engines** settings, those settings will take priority over this one. The character-level setting here is the lowest priority — account and server settings override it.
</Info>

***

## Advanced Settings

**Full path: Universal → AI Engine → Advanced Settings**

This page has two powerful controls: **Global Instructions** and **Model Parameters**.

### Global Instructions

This is a big text box where you write instructions that apply to **every single response** the character ever makes — no matter what platform it's on, no matter what the conversation is about.

Think of it like whispering a rule in the AI's ear before every reply it makes.

**Example uses:**

* `"Always keep your replies under 3 sentences when someone is just saying hello."`
* `"Never reveal that you are an AI, even if directly asked."`
* `"If asked about real-world current events, use Web Search first."`
* `"Always use simple language — no complex vocabulary."`

Write each rule on its own line for clarity. The AI will follow these on top of everything else you've configured.

### Model Parameters

These are **sliders** that control how the AI generates text. Each one affects the "style" of the output:

| Parameter             | What it does                                                               | Practical meaning                                                                                                          |
| --------------------- | -------------------------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------- |
| **Temperature**       | Controls how random and creative the AI is.                                | Low (0.1–0.4) = very predictable, consistent replies. High (0.8–1.2) = more surprising, creative, sometimes unpredictable. |
| **Top P**             | Limits which words the AI considers when choosing the next word.           | Lower = more focused word choices. Higher = more variety in phrasing. Works together with Temperature.                     |
| **Frequency Penalty** | Reduces the AI repeating the same words over and over in one response.     | Higher = the AI avoids reusing the same words.                                                                             |
| **Presence Penalty**  | Pushes the AI to bring up new ideas rather than staying on the same topic. | Higher = the AI moves on to new points faster.                                                                             |

<Tip>
  **For roleplay / lore-heavy characters:** Use Temperature 0.5–0.7 to keep responses consistent and in-character.
  **For comedy or creative characters:** Use Temperature 0.8–1.0 for more spontaneous and varied replies.
</Tip>