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

# Models & Reasoning

> Choosing a model and turning reasoning on or off.

# Models & Reasoning

UncensoredGPT lets you pick which AI "model" you talk to. Different models have different strengths, and some can **reason** (think things through step by step before answering).

This guide explains how to choose a model and how to control the thinking.

## What is a "model"?

A model is the "brain" behind the answers. UncensoredGPT offers several **open-source** models. Some are small and fast, others are large and powerful. The **Pro** plan unlocks the Pro models — including **Qwen3.8 Cybersecurity**, **GLM**, **DeepSeek**, **Kimi** and more. You can see all the public models here: [https://huggingface.co/NeuralNet-Hub](https://huggingface.co/NeuralNet-Hub)

Some models are **vision / multimodal**. That means they can look at images as well as read text, so you can upload a picture and ask questions about it.

## How to choose a model

**1. Find the model selector.**

At the top of the chat window there is a box that shows the name of the current model. It usually sits right above the message box, in the chat header. It may show a name like the model family or an icon.

**2. Open the list.**

Click on it. A list of available models will appear.

**3. Pick one.**

Click the model you want. In most cases it switches immediately, and your next message will use that model.

**4. (Optional) Use more than one.**

You can often click a **+** (plus) button next to the selector to add another model, so a few models answer the same question at the same time. See [Multiple Models at Once](https://docs.uncensoredgpt.ai/interface/multi-model-chats.md).

> **Note on plans:** with the Free plan you may not be able to choose your model yourself. Model selection is included with **Pro**. See [Account & Payments](https://docs.uncensoredgpt.ai/interface/account-and-payments.md).

## What is "reasoning" (thinking)?

Some models, when asked a tricky question, **think before they answer**. This is called reasoning, or "thinking". You may see the model show a little **Thinking...** or **Thought** note before it gives its final answer.

Reasoning is useful for math, logic, planning, and hard questions. But not every question needs it. Some people like it, some do not.

## How to turn reasoning ON

Reasoning is not something the AI turns on for you automatically. **You choose a model that has reasoning.** The model list shows which models can reason.

**1. Open the model selector** at the top of the chat.

**2. Look for a "reasoning" flag or badge.**

In the list, models that can think/reason are marked — look for the word **Reasoning** or a **reasoning** tag next to the model name.

**3. Select that model.**

Pick the model with the reasoning flag. From then on, that model will think before it answers, and you will see its thinking step.

## How to turn reasoning OFF

If the model keeps "thinking" and you do not want that, **switch to a model that does not have the reasoning flag.**

**1. Open the model selector.**

**2. Pick a model without the reasoning tag.**

That model will answer directly, without the extra thinking step.

## The simplest rule to remember

> You don't ask the AI to "reason" or "stop reasoning" as a request. You **pick the right model from the model list.** Models marked with the **reasoning** flag think before answering. Models without it answer directly.

## How to answer common questions

**If the user says: "Stop reasoning" / "Can you reason?" / "Why is it thinking so long?"**

> I can't turn reasoning on or off for myself with a request — it depends on the model you picked. Open the model selector at the top of the chat: choose a model **with** the "reasoning" flag to make me think step by step, or one **without** it to answer directly and faster.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.