xPrivo - Privacy-First AI Chat

xPrivo is an open-source AI chat interface and an alternative to ChatGPT/Perplexity that prioritises privacy and control. Chat with xPrivo or any local or external AI model by adding your own API keys. Organise your conversations in workspaces and keep everything local and private. No account is required! Plug-and-play architecture.

By default, most AI assistants use your chats for training and analysis. Even if you opt out of training, your chats will still be anonymised and used for analytics, so you lose control of your valuable thoughts, business ideas and more. By default, xPrivo is private and log-free. You can stay fully worry-free.

Try it now at xprivo.com - Use our pre-configured xPrivo model. We use trusted EU-based partners and the best model for your requests, such as GPT-OSS, Mistral 3, DeepSeek V3.2 or Llama, all of which have a zero-logging guarantee. You can also add your own AI providers with custom API keys or self-host for complete control.

Hosted Solution Benefits:

• Free Tier: Daily request quota with occasional non-personalized, non-intrusive ads from vetted partners. Web access.

• PRO Upgrade: Unlimited daily requests and ad-free experience

• Privacy-First Authentication: No account creation required. You receive a unique license key that serves as your identifier while keeping all chat history exclusively on your local device

Become a Sponsor / Gain visibility on xPrivo

Support xPrivo to help us keep the service free and reach our user base through carefully selected, non-personalized, non-intrusive advertising. Gain visibility within our community without compromising privacy or user experience.

Learn more about sponsorship / advertising.

Big sponsors:

• Hetzner - Server, cloud & hosting
• Bunny - CDN, DDOS protection, S3 buckets & DNS
• Ionos - Domains & web hosting

Why xPrivo?

xPrivo is designed for users who want a powerful AI chat experience without compromising on privacy:

• Privacy-First: All data stored locally in your browser. No servers, no tracking
• Provider Agnostic: Connect to any OpenAI-compatible API endpoint
• No Account Required: Start chatting immediately with your own API keys
• Full Control: Self-host or use our hosted version at xprivo.com
• Customizable: Brand it and theme it the way you like it

Whether you're a developer, researcher, or privacy-conscious user, xPrivo provides a clean, modern interface for interacting with AI models on your terms.

Features

Core Functionality

• Modern chat interface with real-time streaming responses
• Connect any AI provider: Local models or OpenAI, Anthropic and more if needed
• Dark/light mode with automatic detection
• Multi-language support: English, French, German, Hindi, and more

Advanced Capabilities

• File processing: Upload and discuss PDF, CSV, Excel, and DOC files with @mention system (complex PDFs are still a problem as of now)
• Workspaces: Organize conversations with custom instructions
• AI Experts: Create specialized assistants with custom knowledge
• Web search integration: Real-time information retrieval
• Import/Export: Transfer conversations between devices
• Tone preferences: Personalize AI response styles

User Experience

• Fully responsive design for mobile and desktop
• Fast and lightweight - no bloat and kept minimalistic
• Customizable branding and assistant icons
• Accessible from anywhere via xprivo.com

Quick Start / Installation

Option 1: Use the Hosted Version

Simply visit xprivo.com and start chatting.

You can then also connect to other AI providers:

1. Go to Settings > Configure AI Models & Endpoints
2. Click Add Provider
3. Enter your provider details:
   • Name: A reference name (e.g., "My OpenAI")
   • Endpoint URL: Your API endpoint (e.g., https://api.aiprovider.com/v1/chat/completions)
   • Authorization: Your API key (e.g., Bearer sk-...)
   • Model: The model name (e.g., llama-3.3, deepseek)
4. Start chatting!

Note: When using xprivo.com with certain AI providers, you may need to set up a proxy server to handle CORS. See the Backend Proxy Setup section below.

Option 2: Self-Host

Prerequisites:

• Node.js 18 or higher
• npm (comes with Node.js)

Installation:

# Clone the repository
git clone 
cd xprivo-ai-chat

# Install dependencies
npm install

# Start development server
npm run dev

The app will open at http://localhost:5173

Configure your default AI provider in src/config/setup.ts:

defaultProvider: {
  name: 'Your Provider Name',
  url: 'https://api.yourprovider.com/v1/chat/completions',
  authorization: 'Bearer your-api-key-here',
  model: 'llama-3.3-70b',
  enableWebSearch: false,
  enableSafeWebSearch: true
}

Important: When running locally and calling external endpoints, you may encounter CORS issues with some AI providers. See the Backend Proxy Setup section to resolve this.

Backend Proxy Setup

When Do You Need This?

Some API providers (like Openrouter, etc.) may block direct API requests from your browser due to CORS (Cross-Origin Resource Sharing) restrictions. If you encounter CORS errors, you'll need to proxy requests through your own server. You could set this up with a lightweight Node.js proxy server with an OpenAI-compatible endpoint structure.

Once your "middleman" proxy server is running, you can do the following in the frontend:.

In the chat interface, go to Settings > Configure AI Models & Endpoints and add:

• Name: MyCoolAiProvider (or your preferred name)
• Endpoint URL: http://localhost:3001/chat
• Authorization: (leave empty or use Bearer dummy - auth is handled by proxy)
• Model: llama-3.3 (or your chosen model)

You need to route the models to the correct AI provider in the backend.

Customization

Request Configuration

Modify API request parameters globally in src/config/setup.ts:

apiRequestConfig: {
  temperature: 0.7,
  max_tokens: 2000,
  max_completion_tokens: 4000,
  // ... other parameters
}

Note: These settings apply to all models in your configuration.

Branding Your App

App Name: Edit translations in src/translations/index.ts

Logo: Update menu_icon path in src/config/setup.ts and sidebar logo

Colors: Modify tailwind.config.js for custom color schemes

Custom Assistant Avatar

Change the assistant's profile picture:

• Supports PNG, JPG, JPEG, SVG, WebP
• Persists across sessions
• Auto-fallback to default icon

File Upload Settings

Adjust maximum file size in src/utils/fileProcessor.ts:

const maxSize = 10 * 1024 * 1024; // 10MB default

Adding Languages

Add new languages in src/translations/index.ts by including all required translation keys.

Building for Production

npm run build

The optimized build will be in the dist/ folder, ready for deployment to any static hosting service (Vercel, Netlify, etc.).

Troubleshooting

Chat not working?

• Verify your API endpoint URL is correct and .includes the proper path (e.g., /v1/chat/completions).
• Confirm your API key/bearer token is valid
• Ensure that the model name matches exactly what your provider expects.
• Check the browser console for error messages.
• Test Cross-Origin Resource Sharing (CORS).
• Check whether it is an OpenAI-compatible API.

CORS errors?

See the Backend Proxy Setup section above to proxy requests through your own server.

File uploads failing?

• Check that the files are under the size limit. This is 10 MB, but this may be increased in newer versions, so check for yourself.
• Verify the file type is supported (PDF, CSV, Excel, DOC)

Streaming responses not appearing or getting a "Try again later" error?

• Confirm your API endpoint supports Server-Sent Events (SSE).
• Check for CORS issues in the browser console.
• Verify that your proxy server (if using one) is forwarding streaming responses properly.
• Remove any special parameters sent in the header. (Paramters starting with X-...)

Technology Stack

• React 18 with TypeScript
• Tailwind CSS for styling
• Vite for lightning-fast development
• Lucide React for beautiful icons
• LocalForage for local data persistence

License

This project is licensed under the AGPLv3 License - see the LICENSE file for details.

Support the Project

If you find xPrivo useful, please star this repository and report any bugs or suggest new features via the Issues tab. You can also share it with others who value privacy.

Built with ❤️ for the privacy-conscious community.