Local Unity - AI Chat + Image Generation

By Unity AI Lab | www.unityailab.com

A completely local AI assistant that runs entirely on your computer. No internet required, no data sent anywhere, 100% private.

Features

• Chat with AI - Local LLM text generation (Phi-3, Llama 3, Mistral, Qwen2)
• Generate Images - Stable Diffusion image generation (DreamShaper, Realistic Vision)
• Session Management - Save and organize chat history
• Memory System - Persistent context across sessions
• Customizable - Change the AI's personality with system prompts
• Voice Output - Text-to-speech support
• Dark UI - Hacker-themed interface

Everything runs on YOUR hardware using YOUR graphics card. No subscriptions, no API keys, no cloud services.

System Requirements

Minimum

• OS: Windows 10/11 (64-bit) or Linux
• GPU: NVIDIA GTX 1060 6GB or better (CUDA required)
• RAM: 16 GB
• Storage: 20 GB free space
• Python: 3.10 or newer

Recommended

• GPU: NVIDIA RTX 3060 12GB or better
• RAM: 32 GB
• Storage: 50 GB free space (for multiple models)

Required Software

• Python 3.10+ (check "Add to PATH" during install)
• NVIDIA GPU Drivers (latest version)
• CUDA Toolkit (12.x recommended)

Quick Start

Windows

# Double-click run.bat
run.bat

Linux/Ubuntu

chmod +x start.sh
./start.sh --setup    # First time only - installs system deps
./start.sh            # Start the server

First run will:

1. Create virtual environment
2. Install PyTorch with CUDA
3. Install llama-cpp-python with CUDA
4. Install diffusers for image generation
5. Start server at http://localhost:8080
6. Auto-open browser (Windows only)

How It Works

You type a message
        |
Browser sends it to the server (server.py)
        |
Server sends it to the AI model on your GPU
        |
AI thinks and generates a response
        |
Response appears in your browser

File Structure

LocalUnity/
├── run.bat              # Windows launcher
├── start.sh             # Linux/Ubuntu launcher
├── server.py            # Flask server
├── index.html           # Web interface
├── data/
│   ├── system_prompt.md # AI personality
│   ├── memories.json    # Persistent memory
│   └── sessions.json    # Chat sessions
├── docs/
│   ├── README.md        # User documentation
│   └── README_TECHNICAL.md  # API reference
└── models/              # AI models (auto-downloaded)
    ├── llama3/          # Text models (GGUF)
    ├── mistral/
    ├── phi3/
    └── image/           # Image models (Diffusers)

Adding Models

Text Models (GGUF format)

Place .gguf files in models/ folder:

models/
├── llama3/Llama-3.2-3B-Instruct-Q4_K_M.gguf
├── mistral/Mistral-7B-Instruct-v0.3.Q4_K_M.gguf
└── phi3/Phi-3-mini-4k-instruct-q4.gguf

Image Models (Diffusers format)

Place model folders in models/image/:

models/image/
├── dreamshaper-8/
├── realistic-vision-v51/
└── epicrealism/

API Reference

See docs/README_TECHNICAL.md for full API documentation.

Quick API Examples

Chat:

POST /api/chat
{"message": "Hello", "session_id": "uuid"}

List Models:

GET /api/models
GET /api/image/models

Generate Image:

POST /api/image/generate
{"prompt": "a cat in space", "width": 512, "height": 512}

Contact

• Website: www.unityailab.com
• Email: unityailabcontact@gmail.com
• Creators: Hackall360, Sponge, GFourteen

License

Private project by Unity AI Lab.