Browser Use

What is Browser Use?

Browser Use is an open source Python library and cloud platform for AI agents that control a web browser. It is built by Browser Use and lets a model complete tasks such as filling forms, collecting data and navigating sites by reading pages and taking actions in a real browser. The problem it solves is automation for websites that have no usable API: instead of writing brittle selectors for every step, teams describe the task and let an agent work through the page.

Key features

• Natural language tasks. An agent receives a task in plain language and decides which links, fields and buttons to use as it works through a site.
• Model choice. The library supports hosted models from OpenAI, Anthropic and Google, along with local and custom model providers.
• Open source core. The Python package is open source, so teams can inspect the agent loop and run it on their own infrastructure.
• Managed cloud browsers. The cloud product provides hosted browsers, stealth browser settings and residential proxies for sites that block data center traffic.
• Web agent API. A hosted API runs browser tasks without a local browser.
• MCP and coding agent integrations. Browser Use ships an MCP server and integrations for coding agents such as Claude Code, Codex, Cursor and Gemini.

Who is Browser Use for?

Developers and automation teams that need to interact with websites that lack APIs. It suits AI engineers building agents that research, shop, book or fill forms, data teams that collect information from sites with login walls or dynamic pages, QA teams testing real user journeys, and operations teams automating repetitive browser work.

Use cases

• Collecting product, pricing or listing data from sites without a public API.
• Filling and submitting forms across supplier or government portals.
• Testing a sign-up and checkout flow as a real user would.
• Researching a company by reading several pages and summarising the findings.

Pricing

Browser Use Cloud is pay as you go with no subscription. Cloud browsers cost $0.02 per hour plus $5 per GB for residential proxy traffic. Web agents are billed at the model price plus 20 percent, plus browser time and traffic. Concurrency rises with total payments: new projects start with 10 concurrent browsers, a $100 top-up unlocks 50, $1,000 unlocks 250, $5,000 unlocks 500 and $25,000 unlocks 1,000. See the official Browser Use pricing page.

What's new

The latest open source release, version 0.13.14, was published on 26 October 2026 and is described as Python agent and integration updates. Read the 0.13.14 release notes.

Getting started

1. Install the package and the Chromium browser it needs.
2. Add an API key for Browser Use Cloud or another supported model provider to a .env file.

pip install uv
uv venv --python 3.12
source .venv/bin/activate
uv pip install browser-use
uvx browser-use install

from browser_use import Agent, ChatBrowserUse
import asyncio
 
async def main():
    agent = Agent(task="Find the top post on Hacker News", llm=ChatBrowserUse())
    await agent.run()
 
asyncio.run(main())

Alternatives

• Browserbase provides managed headless browsers for agents and automation.
• Browserless offers hosted browser infrastructure for Playwright and Puppeteer.
• Firecrawl turns websites into clean markdown for AI applications.
• Playwright is an open source browser automation library for scripted tests.

FAQ

Is Browser Use free?

The open source Python library is free to use. The cloud platform is pay as you go, with $1 in free credit for new accounts and charges for browser time, proxy traffic and model usage.

What is Browser Use used for?

Browser Use is used to automate browser tasks with AI agents, such as filling forms, extracting data, testing user journeys and researching websites that do not offer an API.

Does Browser Use work with OpenAI models?

Yes. It supports OpenAI, Anthropic and Google models, plus local and custom model providers.

Browser Use vs Playwright?

Playwright runs scripted browser steps that a developer writes in advance. Browser Use gives an AI agent a task and lets it decide the steps at run time. Playwright is better for stable regression tests, while Browser Use is better for open-ended tasks and sites that change often.

Yes. The open source library runs on your own machine or infrastructure. Managed browsers, proxies and the hosted agent API are part of the paid cloud product, and on-premises deployment is available through enterprise sales.