Local Chat AI 🤖

A powerful, privacy-focused, and completely offline AI chat application built with Flutter. Run state-of-the-art Large Language Models (LLMs) like Llama 3 and Qwen directly on your device—no internet connection required!

Created by Haris.

🌟 Key Features

• 100% Offline Inference: Chat securely without your data ever leaving your device.
• Built-in Model Downloader: Easily browse, download, and manage advanced open-source models (GGUF format).
• Curated Model Selection: Includes optimized models like:
  • Llama 3.2 (1B and 3B Instruct)
  • Qwen 3.5 (0.8B to 4B variants)
  • Qwen 2.5 (Abliterated/Uncensored models)
• Real-time Performance: fluid typing indicators and fast token generation.
• Modern UI/UX: Clean, intuitive interface inspired by top-tier chat applications.
• Chat History Persistence: Automatically saves your conversations for when you return, using local storage.
• Resource Management: View download speeds, progress, and easily delete models to free up space.

🛠️ Technology Stack

• Frontend Framework: Flutter / Dart Focus on a unified codebase for Android and cross-platform compilation.
• State Management: MultiProvider (Provider Pattern) + MVVM Architecture for robust data binding.
• Local AI Inference: Powered by llama_cpp_dart for high-performance GGUF model execution.
• Local Storage: shared_preferences for quick app settings and chat persistence.
• Network (for Downloads): dio for reliable, resumable file downloading of large LLM files.
• File System: path_provider for managing downloaded models securely on the device.

🚀 Getting Started

Prerequisites

• Flutter SDK (3.x or higher)
• Android Studio / Xcode for emulators or physical device testing
• Recommended Device: A modern smartphone with at least 4GB of RAM (6GB-8GB recommended for larger models).

Installation

1. Clone the repository:

   git clone https://github.com/1haris2/local_chat_ai.git
   cd local_chat_ai

2. Install dependencies:

   flutter pub get

3. Run the app:

   flutter run

   Note: For the best inference performance, always test on a physical release build rather than an emulator in debug mode.

Building for Production (APK)

To generate a highly optimized and secure release Android APK with code obfuscation (to prevent reverse engineering):

flutter build apk --release --obfuscate --split-debug-info=build/app/outputs/symbols

🛡️ Security & Privacy

• Encrypted Storage: All chat histories and sessions are stored using FlutterSecureStorage, which uses AES encryption backed by the device's hardware-level keychain (Android Keystore).
• 100% Local Inference: AI models run entirely on-device. No text or data is ever sent to a server.
• Manifest Hardening: The app follows the principle of least privilege, minimizing the permissions requested in the Android manifest.

📱 Screenshots & Previews

(Feel free to add screenshots of your App's Model Selector, Chat Interface, and Settings here!)

🤝 Contributing

Contributions, issues, and feature requests are welcome! Feel free to check out the issues page.

📄 License

This project is open-source and available under the MIT License.

---

*Developed with ❤️ by Haris*