ry-chat

Ry Chat (Multi-Provider LLM Chat Interface 💬)

A versatile, lightweight Streamlit web application that provides a unified chat interface for OpenAI, NVIDIA NIM, and local Ollama language models. It features real-time response streaming, image/document upload capabilities, and live performance metrics (latency and tokens per second).

📸 Demo

🚀 Features

• Multi-Provider Support: Seamlessly switch between OpenAI, NVIDIA NIM, and local Ollama instances.
• Dynamic Configurations: Automatically swaps default endpoints, API key parameters, and recommended models (gpt-5, qwen2.5, etc.) based on your selection.
• Multimodal Ready: Supports uploading images (png, jpg, jpeg) and documents (pdf, txt) directly into the conversation.
• Real-time Streaming: Smooth, token-by-token response rendering.
• Performance Metrics: Tracks and preserves generation latency ($seconds$) and processing speed ($tokens/sec$) in the chat history.

---

📂 Project Structure

├── .env                  # Your local environment secrets (ignored by git)
├── .env.example          # Template for environment variables
├── .gitignore            # Git ignore configurations
├── main.py               # Main Streamlit application source code
└── requirements.txt      # Python dependencies

---

🛠️ Setup Instructions

1. Clone the Repository

Navigate to your project directory containing these files.

2. Create a Virtual Environment & Install Dependencies

It is highly recommended to use a virtual environment to manage your dependencies.

# Create a virtual environment
python -m venv venv

# Activate the virtual environment
# On Windows (PowerShell):
.\venv\Scripts\Activate.ps1
# On macOS/Linux:
source venv/bin/activate

# Install required packages
pip install -r requirements.txt

Note: Ensure your requirements.txt includes at least: streamlit, openai, python-dotenv.

3. Configure Environment Variables

Copy the provided .env.example file to create your local configurations:

cp .env.example .env

Open the newly created .env file and populate it with your respective API credentials:

OPENAI_API_KEY=your_openai_api_key_here
NVIDIA_API_KEY=your_nvidia_api_key_here
OLLAMA_API_KEY=ollama

---

💻 Running the Application

To launch the web interface, execute the following command in your terminal:

streamlit run main.py

Once running, your default browser should automatically open to http://localhost:8501.

---

💡 How to Use

1. Select a Provider: Use the dropdown menu in the sidebar to choose between Ollama, OpenAI, or NVIDIA.
2. Review/Edit Endpoints: The app will auto-fill the default Base URL and API key from your .env file. You can override these directly from the UI if needed.
3. Attach Files: Drop an image or text document into the file uploader if you want to use vision or grounding capabilities (ensure the chosen model supports multimodal inputs).
4. Chat: Type your prompt into the input bar at the bottom and hit Enter!