# Owl Browser **Stealth Browser Automation Built for AI Agents.** Replace screenshot-heavy agent loops with Agent Rendering: compact, structured page views with addressable controls, running on a stealth, self-hosted Chromium engine. ## Key Features - **Agent Rendering (v1.2.0)**: Compact, structured page views with addressable controls. - **Source-Level Stealth**: Real fingerprints originating from native C++ code paths. - **185 Automation Tools**: Comprehensive set of tools for browser control. - **Model Context Protocol (MCP)**: Built-in HTTP transporter MCP server. - **AI Vision Model**: Local CAPTCHA solving with an on-device model. - **Multi-Context Architecture**: Supports up to 256 parallel contexts. - **Docker Deployment**: Self-hosted container deployment. - **Developer SDKs**: First-class support for Python and Node.js. ## Benchmarks In our latest agentic benchmarks, Owl Browser demonstrated exceptional token efficiency: - **Tokens per Task**: ~38.6k tokens - **Efficiency**: Roughly half the tokens compared to alternative stacks (~71.4k and ~73.4k tokens) while maintaining the same correctness. ## Pricing - **Developer**: $49.99/mo (Cloud sandbox for evaluation, 1 seat, 5 active contexts) - **Starter**: $1,999/mo (Scale on your own infrastructure, 3 seats, unlimited active contexts) - **Business**: $19,999 once + $3,999/mo support (Priority support, 10 seats, custom integrations) - **Enterprise**: $49,999 once + $9,999/mo support (Maximum scale with white-label options, 50 seats) ## Quick Links - [Documentation](https://owlbrowser.net/docs) - [Python SDK](https://owlbrowser.net/python-sdk) - [Node SDK](https://owlbrowser.net/node-sdk) - [Benchmarks](https://owlbrowser.net/benchmark) - [FAQ](https://owlbrowser.net/faq) - [GitHub Organization](https://github.com/Olib-AI) - [Owl Light (Free Browser) Repo](https://github.com/Olib-AI/owl-light)