Browser-Use API - Self-Hosted AI Browser Automation by Ananiya KestoBrowser-Use API - Self-Hosted AI Browser Automation by Ananiya Kesto

Browser-Use API - Self-Hosted AI Browser Automation

Ananiya Kesto

Completed work

AI Automation

Docker

Docker Compose

Python

Computer Software

🚀 Browser-Use REST API - Watch Your AI Work in Real-Time

The Most Advanced Self-Hosted Browser Automation API

Drop-in replacement for Browser-Use Cloud with real-time VNC viewing, browser session management, and 15+ LLM models

🎯 Quick Start · 📖 Documentation · 🐳 Docker Hub · 💬 Discord

✨ What Makes This Special?

👁️ Watch AI Work in Real-Time

See exactly what your AI agent is doing through browser VNC streaming. Debug faster, understand better.

🔄 Persistent Browser Sessions

Create browser profiles, save login states, and reuse sessions across multiple tasks - just like a human would.

🎮 Full Task Control

Start, stop, pause, resume tasks. Get real-time step updates as your agent executes.

🌐 15+ LLM Models

Switch between GPT-4, Claude, Gemini, Llama and more. Get $10 free with browser-use-llm.

🔓 100% Self-Hosted

Complete control over your data. No cloud dependencies. Works offline.

⚡ Production Ready

Docker support, multi-arch builds, health checks, supervisor process management, and CI/CD ready.

🎬 See It In Action

Output:

Navigate to http://localhost:8080/vnc.html and watch your AI agent work in real-time! 🎥

🚀 Quick Start (3 Commands)

Option 1: Docker (Recommended - 30 seconds)

✅ Done! API is running at http://localhost:8080

👁️ VNC Viewer: http://localhost:8080/vnc.html 📚 API Docs: http://localhost:8080/docs

Option 2: Docker Compose (For Development)

Option 3: Local Python (Advanced)

🎯 Real-World Use Cases

🔍 Research Automation

Perfect for researchers, analysts

🛒 E-commerce Monitoring

Price tracking, competitor analysis

📊 Data Collection

Web scraping, lead generation

🎫 Booking & Reservations

Restaurant bookings, event tickets

🧪 QA Testing

E2E testing, regression testing

📱 Social Media

Social media automation

🌟 Key Features

🎥 Real-Time Browser Viewing (VNC)

NEW! Watch your AI agent work through a browser interface. No more blind automation.

Features:

🖥️ Full HD 1920x1080 display

🔄 Real-time screen updates

🖱️ Interactive mode (click, type, scroll)

📹 Watch task execution live

🐛 Debug issues instantly

🔄 Browser Session Management

Create persistent browser sessions with saved cookies, localStorage, and authentication:

Features:

💾 Save/load browser profiles

🍪 Persistent cookies and sessions

🔐 Keep login states across tasks

📁 Share profiles between tasks

♻️ Reuse authenticated sessions

📊 Advanced Task Tracking

Get real-time step-by-step updates as your agent executes:

🤖 15+ LLM Models Supported

Provider Models API Key Required Browser-Use browser-use-llm (Get $10 free) ✅ Get Key OpenAI gpt-4.1, gpt-4.1-mini, gpt-4o, gpt-4o-mini, o4-mini, o3 ✅ Google gemini-2.5-flash, gemini-2.5-pro, gemini-flash-latest, gemini-flash-lite-latest ✅ Anthropic claude-sonnet-4, claude-sonnet-4-5, claude-3-7-sonnet ✅ Meta llama-4-maverick-17b-128e-instruct ✅

💡 Tip: Start with gemini-flash-lite-latest (fast & cheap) or browser-use-llm ($10 free credit).

📖 API Endpoints

🎯 Task Management

Endpoint Method Description /tasks POST Create a new browser automation task /tasks GET List all tasks (with pagination & filters) /tasks/{id} GET Get detailed task info with steps /tasks/{id} PATCH Control task (stop, pause, resume) /tasks/{id}/vnc GET Get VNC viewer URL for task /tasks/{id}/logs GET Download task execution logs /tasks/{id}/debug GET Debug endpoint for raw data

🌐 Browser Session Management (NEW!)

Endpoint Method Description /browser/launch POST Launch a browser with profile /browser/sessions GET List all active browser sessions /browser/{id} DELETE Close a browser session /browser/{id}/navigate POST Navigate browser to URL /browser/profiles GET List saved browser profiles /browser/profiles/{name} DELETE Delete a browser profile

🔧 System & Health

Endpoint Method Description /health GET API health check /vnc/health GET VNC services status / GET API info & version /docs GET Interactive Swagger UI

🔥 Advanced Examples

Example 1: Multi-Step Research with VNC Viewing

Example 2: Persistent Login Session

Example 3: E-commerce Price Monitoring

Example 4: Parallel Task Execution

🐳 Production Deployment

Docker Compose (Recommended)

Environment Variables

Scaling with Kubernetes

🆚 Self-Hosted vs Cloud

Feature Self-Hosted (This) Browser-Use Cloud Cost 🟢 Free (your hardware) 🟡 Pay per task Privacy 🟢 100% private 🟡 Data sent to cloud VNC Viewing 🟢 Real-time browser view 🔴 Not available LLM Models 🟢 15+ models 🟢 15+ models API Compatibility 🟢 v2 compatible 🟢 v2 native Offline Mode 🟢 Works offline 🔴 Requires internet Setup Time 🟡 5 minutes 🟢 Instant Scalability 🟡 Manual scaling 🟢 Auto-scaling Support 🟡 Community 🟢 Official support

💡 Best of both worlds: Use self-hosted for development/testing, cloud for production at scale!

🔧 Configuration & Customization

Request Schema

Response Schema

🐛 Debugging & Troubleshooting

Check Service Health

View Container Logs

Debug Task Execution

Common Issues

VNC not connecting

Browser crashes/hangs

LLM API key errors

📊 Performance Benchmarks

Task Type Avg Time Success Rate Best LLM Simple navigation 5-10s 99% gemini-flash-lite Form filling 15-30s 95% gpt-4o-mini Data extraction 30-60s 92% gpt-4.1 Complex workflows 2-5min 85% claude-sonnet-4 Multi-step research 5-10min 80% o3

Tested on: 4-core CPU, 8GB RAM, 50Mbps internet