Quorum

Instead of asking one LLM, convene a Quorum. Multiple models answer your question independently, critique each other anonymously, then a chairman synthesizes the final answer.

Fully local, lightweight web app. Minimal, readable code. Runs on localhost.

How It Works

Each query goes through 3 stages:

1. Stage 1 — Individual Responses: All council models answer independently. Tab view lets you inspect each response.
2. Stage 2 — Peer Rankings: Each model evaluates the others' responses. Identities are anonymized (Response A, B, C…) to prevent bias. Each model ranks the rest by accuracy and insight.
3. Stage 3 — Final Synthesis: The Chairman model (strongest) reads all responses + peer rankings and compiles the definitive answer.

Features

• Dark mode UI with progressive streaming (each stage appears as it completes)
• Sidebar with conversation history — create, rename, and delete chats
• Transparent: inspect raw model outputs, extracted rankings, and aggregate scores
• Persistent conversations (JSON storage, no database needed)
• Single OpenRouter API key covers all model providers

Setup

1. Get an OpenRouter API Key

Sign up at openrouter.ai and create an API key. Create a .env file in the project root:

OPENROUTER_API_KEY=sk-or-v1-...

Never commit this file. It is already in .gitignore.

2. Install Dependencies

Backend (requires Python 3.10+):

With uv (recommended):

uv sync

Or with pip:

pip install fastapi uvicorn python-dotenv httpx pydantic

Frontend:

cd frontend
npm install

3. Configure Models (Optional)

Edit backend/config.py to customize the council:

COUNCIL_MODELS = [
    "openai/gpt-4o-mini",
    "google/gemini-2.5-flash",
    "anthropic/claude-haiku-4-5",
    "meta-llama/llama-3.3-70b-instruct",
]

CHAIRMAN_MODEL = "google/gemini-2.5-flash"

Any model available on OpenRouter works here.

Running

Option 1: Start script

./start.sh

Option 2: Manual

Terminal 1 — Backend:

uv run python -m backend.main
# or: python -m backend.main

Terminal 2 — Frontend:

cd frontend
npm run dev

Open http://localhost:5173 in your browser.

Tech Stack

Layer	Technology
Backend	FastAPI (Python 3.10+), async httpx
Frontend	React 19 + Vite, react-markdown, SweetAlert2
Models	OpenRouter API (single key, all providers)
Storage	JSON files in data/conversations/
Package mgmt	uv (Python), npm (JS)

Ports

• Backend: http://localhost:8001
• Frontend: http://localhost:5173