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

# Private Mode

> Complete privacy on Cocoon's decentralised GPU network

## Overview

For users who value freedom and control, Mira offers **Private Mode** — where requests are processed on [Cocoon](https://cocoon.network)'s decentralised GPU network with zero data sharing.

## How it works

<Steps>
  <Step title="User enables Private Mode">
    Open the Mira mini-app → tap *Discover* → enable Private Mode.
  </Step>

  <Step title="Request goes to Cocoon">
    Instead of OpenAI's servers, your message is routed to a Cocoon TEE node.
  </Step>

  <Step title="Processed in TEE">
    The request is processed inside a **Trusted Execution Environment** — hardware-level isolation that prevents even the node operator from reading your data.
  </Step>

  <Step title="Response returned">
    The result comes back to you in Telegram. Nothing is stored.
  </Step>
</Steps>

## What is Cocoon?

**Cocoon** (Confidential Compute Open Network) is a decentralised network of confidential GPU computing, integrated with Telegram and TON.

* Nodes run in **TEE** (Trusted Execution Environment)
* Data is **not stored** on external servers
* **Not shared** with any third parties
* Powered by **Qwen3-32B** model

## Normal vs Private Mode

| | Normal Mode | Private Mode |
| - | - | - |
| Model | GPT-4o (OpenAI) | Qwen3-32B (Cocoon) |
| Data storage | OpenAI infrastructure | Not stored |
| Privacy | Standard | Maximum |
| Speed | Fast | Fast |
| Cost | Standard tokens | Standard tokens |

## Enable Private Mode

<Card title="Open Mira Mini-App" icon="shield" href="https://t.me/mira/home" horizontal>
  Tap *Discover* at the bottom of the chat to switch to Private Mode.
</Card>


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