The General-Purpose Agentic Browser
Give your AI agent a real browser. 2-6x fewer tokens than every competitor, tested across MCP and CLI benchmarks with 100% accuracy.
openbrowser.me
Give your AI agent a real browser. 2-6x fewer tokens than every competitor, tested across MCP and CLI benchmarks with 100% accuracy.
Built different from the ground up.
Tested against 6 competitors across 2 independent benchmarks. Text-first architecture means 2-6x fewer tokens, every time.
Works with 15 LLM providers out of the box, plus any OpenAI-compatible endpoint.
One execute_code tool, persistent Python namespace. Navigate, click, type, extract.
Watch your agent browse in real-time via VNC streaming. See every click, scroll, and navigation live.
Docker, Kubernetes, cloud deployment. Battle-tested infrastructure for any scale.
MIT licensed. Community-driven. Fully extensible. Build on top of OpenBrowser.
Star on GitHubFrom zero to autonomous browsing in minutes.
The agent navigates, clicks, types, and extracts data autonomously. Watch it work in real-time through the live browser view.
From ad-hoc browsing to hands-off scheduled automation.
Watch a real agent browse the web autonomously.
Schedule recurring browser tasks that reuse saved login sessions.
OpenBrowser integrates with any MCP-compatible client.
Two research projects developing custom fine-tuned models, with open-source weights on HuggingFace.
SFT + GRPO on Qwen3-8B with browser execution rewards across 1,250 tasks spanning 8 domains including healthcare, finance, and legal.
First cross-paradigm study of diffusion LMs for web planning. MDPO achieves near-parity with autoregressive models on form-filling benchmarks.
Trained weights available on HuggingFace, ready for inference or further fine-tuning.
Be the first to try the hosted version of OpenBrowser.