Skip to content

kalil0321/the-browser-arena

Repository files navigation

The Browser Arena

Compare AI browser automation agents side-by-side. Submit a task, watch agents run in real-time, and compare speed, cost, and success.

Arena Demo

What it does

  • Multi-agent runs: Browser-Use, Smooth, Stagehand (local & cloud)
  • Live comparison: Parallel execution with browser views
  • Metrics built-in: Time, steps, cost, success
  • History & replay: Sessions are saved automatically
  • Multi-LLM: OpenAI, Gemini, Claude

Quick start

  1. Install and run the web app
npm install
npm run dev
  1. Start the agent server (Python)
cd agents
source .venv/bin/activate
uv sync        # or: pip install -r requirements.txt
python server.py
  1. Create .env.local
# Convex
NEXT_PUBLIC_CONVEX_URL=https://your-deployment.convex.cloud
CONVEX_DEPLOYMENT=your-deployment

# LLMs (add at least one)
OPENAI_API_KEY=...
GOOGLE_API_KEY=...
ANTHROPIC_API_KEY=...
BROWSER_USE_API_KEY=...

# Optional services
SMOOTH_API_KEY=...
BROWSERBASE_API_KEY=...
BROWSERBASE_PROJECT_ID=...

# Agent server
AGENT_SERVER_URL=http://localhost:8080
STAGEHAND_SERVER_URL=http://localhost:3001
  1. Convex (first time)
npx convex auth
npx convex dev

Open http://localhost:3000

About

A simple way to test different browser agents with multiple models

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 3

  •  
  •  
  •