Hiyo

Local Intelligence Powered by You

---

Overview

Hiyo is your personal gateway to local AI — a sleek macOS app that brings powerful language models right to your desktop, no cloud required. Built on Apple's native MLX framework, Hiyo delivers exceptional performance on Apple Silicon while keeping everything private and on-device.

Whether you're a developer, researcher, or just curious about AI, Hiyo makes it easy to explore complex ideas with clarity and control. No logins, no data sharing, no internet connection — just fast, focused intelligence that works the way you do.

With Hiyo, you're not just using AI — you're shaping it.

---

✨ Why Hiyo?

Feature	Benefit
🚀 Apple Silicon Native	Built on MLX for up to 2x faster inference vs CPU
🧠 Unified Memory	GPU and CPU share memory — no data copying overhead
🔒 100% Private	Everything runs locally. Zero network calls. Zero data sharing.
⚡ Instant On	No servers to start. Models load directly into memory.
🎯 Curated Models	Pre-configured with optimized 4-bit quantized models
🛡️ Enterprise Security	Sandboxed, hardened runtime, code-signed, encrypted storage

---

System Requirements

Requirement	Details
macOS	14.0 (Sonoma) or later
Chip	Apple Silicon (M1/M2/M3) or Intel
RAM	8 GB minimum, 16 GB recommended
Storage	2 GB for app, 2-8 GB per model
Xcode	15.0+ (for building from source)

Note: MLX-optimized models run best on Apple Silicon. Intel Macs supported with reduced performance.

---

🚀 Quick Start

Option 1: Download Pre-built DMG

1. Download Hiyo-1.0.0.dmg from Releases
2. Open DMG and drag Hiyo to Applications
3. Launch Hiyo from Applications folder
4. On first launch, select a model to download (1-4 GB)
5. Start chatting!

Option 2: Build from Source

See BUILD_APP.md for detailed instructions.

# Quick build
git clone https://github.com/hiyoai/hiyo.git
cd hiyo
swift build -c release

---

🏗️ Architecture

Hiyo is built on a modern, secure architecture:

graph TD
  subgraph "Hiyo - SwiftUI"
    ChatUI["Chat UI"]
    SidebarUI["Sidebar UI"]
    SettingsUI["Settings UI"]
  end

  ChatUI --> Store["HiyoStore - SwiftData"]
  SidebarUI --> Store
  SettingsUI --> Store

  Store --> Provider["MLXProvider - MLX Swift"]
  Provider --> Framework["Apple MLX Framework"]
  Framework --> GPU["Apple Silicon GPU"]

Loading

---

🧠 Included Models

Hiyo comes pre-configured with these optimized models:

Model	Size	Best For
Llama 3.2 1B	0.7 GB	Ultra-fast responses, low memory
Llama 3.2 3B ⭐	1.9 GB	Best balance of speed and quality
Mistral 7B	4.1 GB	Advanced reasoning, complex tasks
Phi-3 Mini	1.8 GB	Efficient, strong performance
Qwen 2.5 7B	4.2 GB	Multilingual, coding
CodeLlama 7B	4.1 GB	Code generation, technical tasks

All models are 4-bit quantized from the MLX Community on Hugging Face.

---

🔒 Security & Privacy

Hiyo implements defense-in-depth security:

Layer	Implementation
Sandbox	Strict macOS App Sandbox
Code Signing	Hardened Runtime with integrity checks
Network	Zero network connections — fully offline
Storage	AES-256 encrypted local database
Memory	Secure memory with automatic wiping
Input	Strict validation and sanitization

Privacy Guarantees

• ✅ No cloud processing
• ✅ No telemetry or analytics
• ✅ No user accounts or tracking
• ✅ No iCloud sync
• ✅ Models cached locally only

---

🛠️ Development

Project Structure

Hiyo/
├── Package.swift                 # Swift Package Manager manifest
├── Sources/Hiyo/
│   ├── HiyoApp.swift            # App entry point
│   ├── ContentView.swift        # Main UI layout
│   ├── HiyoState.swift          # Global state
│   ├── Core/                    # Business logic
│   │   ├── HiyoStore.swift      # Data persistence
│   │   ├── MLXProvider.swift    # MLX integration
│   │   ├── MLXModelRegistry.swift
│   │   └── Models/              # Chat, Message
│   ├── Security/                # 9 security modules
│   ├── UI/                      # SwiftUI views
│   └── Utils/                   # Extensions, constants
├── Tests/                       # Unit & UI tests
└── Assets/                      # Icons, images

Key Technologies

• MLX Swift — Apple's machine learning framework
• SwiftData — Modern persistence
• SwiftUI — Native UI framework
• CryptoKit — Encryption

Building

# Clone
git clone https://github.com/hiyoai/hiyo.git
cd hiyo

# Resolve dependencies
swift package resolve

# Build debug
swift build

# Build release
swift build -c release

# Run tests
swift test

# Create Xcode project
swift package generate-xcodeproj

See BUILD_APP.md for complete DMG creation guide.

---

🎮 Usage

Keyboard Shortcuts

Shortcut	Action
⌘ N	New conversation
⌘ ⇧ N	New window
⌘ ⇧ K	Clear current conversation
⌘ E	Export conversation
⌘ ,	Settings
⌘ ⌫	Delete selected conversation
⌘ ↵	Send message

Tips

• First launch: Download a smaller model (1B or 3B) to test
• Memory pressure: Unload unused models in Settings
• Long conversations: Clear history periodically to free context window
• Export: Conversations export as encrypted .hiyo files

---

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md.

# Fork and clone
git clone https://github.com/yourusername/hiyo.git

# Create branch
git checkout -b feature/amazing-feature

# Make changes, test, commit
swift test
git commit -m "Add: amazing feature"

# Push and open PR
git push origin feature/amazing-feature

Areas for Contribution

• Additional MLX model support
• Conversation search
• Markdown rendering improvements
• Voice input/output
• Plugin system
• Localization

---

📚 Documentation

• BUILD_APP.md — Complete build instructions
• CONTRIBUTING.md — Contribution guidelines
• SECURITY.md — Security architecture details
• API.md — Internal API documentation

---

🙏 Acknowledgments

• **MLX Swift