🧩 LLMs-Assemble App

A Gradio-based web application that lets you query and compare responses from multiple LLM providers simultaneously, all in a single interface.

Designed for side-by-side evaluation, qualitative analysis, and prompt experimentation across models.

---

🖼️ App Screenshot

---

✨ Features

• Multi-LLM Comparison: Query OpenAI, Anthropic, Google, Groq, and Perplexity in parallel.
• Chairman Mode: Enable models to critique, discuss, and respond to each other’s outputs.
• Real-Time Streaming: View responses as they arrive.
• Secure by Default: API keys are stored in-memory only.
• Developer-Friendly: Ideal for prompt testing and model evaluation workflows.

---

🛠️ Tech Stack

• UI: Gradio
• Backend: Python
• LLM Providers:
  • OpenAI
  • Anthropic
  • Google
  • Groq
  • Perplexity

---

📦 Installation

Install the required dependencies:

pip install -r requirements.txt

---

▶️ Usage

Run the application locally:

python app.py

Once started, Gradio will provide a local URL:

Running on local URL: http://127.0.0.1:7860

Open the link in your browser to access the app.

---

🔐 Environment Variables

You can provide API keys either via the UI or by creating a .env file.

OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GOOGLE_API_KEY=AIza...
GROQ_API_KEY=gsk_...
PERPLEXITY_API_KEY=pplx-...

---

📝 Notes

• API keys entered via the UI are stored in-memory only and are cleared when the app restarts.
• For persistent configuration, use environment variables instead.

---

🚧 Project Status

This project is experimental and under active development.

It is inspired by Andrej Karpathy’s llm-council

Expect rough edges and ongoing improvements.

---

📄 License

The code in this repository is licensed under the MIT License.