Universe - Multi-Model Chat Interface

Universe is a versatile web application that provides a unified interface to interact with multiple AI language models from OpenAI and Anthropic, including GPT-4o, Claude 3.7 Sonnet, and more. This application allows users to have persistent conversations with different AI models and seamlessly switch between them.

Features

• Multi-model support: Access to various AI models from OpenAI and Anthropic
  • OpenAI models: GPT-4o, GPT-4o mini, O1, O3-mini
  • Anthropic models: Claude 3.5 Sonnet, Claude 3.7 Sonnet
• Persistent chat history: All conversations are saved to a JSON file and persist across server restarts
• Multiple chat sessions: Create and manage multiple conversations
• Session management: Each user gets their own set of conversations
• Streaming responses: Real-time streaming for supported models
• Clean, simple interface: Elegant and responsive UI with dark/light mode
• Keyboard shortcuts: Improve productivity with keyboard navigation

Installation

1. Clone the repository:

git clone https://github.com/yourusername/universe.git
cd universe

2. Create a virtual environment and activate it:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install the required dependencies:

pip install flask openai anthropic python-dotenv

4. Create a .env file in the project root with your API keys:

OPENAI_API_KEY=your_openai_api_key
CLAUDE_API_KEY=your_anthropic_api_key
FLASK_SECRET_KEY=your_secret_key_for_flask_sessions

Usage

1. Start the Flask server:

python app.py

2. Open your web browser and navigate to:

http://localhost:3000

3. Start chatting with any of the available AI models by selecting them from the dropdown menu.

File Structure

• app.py: Main Flask application
• templates/index.html: Main HTML template for the web interface
• chat_histories.json: Persistent storage for all chat histories

Key Features Explained

Chat History Persistence

All chat histories are automatically saved to a JSON file (chat_histories.json) after every interaction. This ensures that conversations persist even if the server is restarted. The chat histories are organized by user ID and chat ID.

Message History Limit

By default, the application only sends the last 5 messages from the conversation history to the AI model. This helps control token usage and response times. If you want to change this limit, modify the MESSAGE_HISTORY_LIMIT variable in app.py:

# Change this value to increase or decrease the number of messages sent to the AI model
MESSAGE_HISTORY_LIMIT = 5  

Increasing this value will provide more context to the AI model but may increase API costs and response times. Decreasing it will reduce context but may save on costs.

Model Selection

Users can select different AI models from the dropdown menu. Each model has its own configuration, including provider (OpenAI or Anthropic) and streaming capability.

Multiple Chat Sessions

Users can create multiple chat sessions and switch between them. Each chat session has its own history and can use different AI models.

Keyboard Shortcuts

• / (Forward slash): Focus the input field
• Alt+Enter: Focus the input field
• Ctrl+N or Cmd+N: Create a new chat
• Shift+Cmd+S: Toggle sidebar
• Ctrl+D or Cmd+D: Toggle dark/light mode
• Arrow keys for message navigation

Customization

System Prompt

You can customize the system prompt in the app.py file. The system prompt defines the AI assistant's behavior and is sent with every request to the AI models.

basic_system_prompt = '''
You are AI coding Assistant. Imagine you are a GOD of programming
and you can do basically everything user request to do.
Answer shortly always, if not requested for long response.
'''

Adding New Models

To add new models, update the MODELS dictionary in the app.py file:

MODELS = {
    'new-model': {
        'provider': 'openai',  # or 'anthropic'
        'name': 'actual-model-name', 
        'stream': True  # or False
    },
    # ...
}

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

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

Acknowledgements

• OpenAI for providing the GPT models
• Anthropic for providing the Claude models
• Flask for the web framework