---
title: "BrowserOS"
description: "The open-source AI browser that turns your words into actions"
---

<Frame>
  <img src="/images/hero-dark.png" alt="BrowserOS in action" />
</Frame>

BrowserOS is an AI-native browser that lets you automate any web task by describing it in plain English. Built on Chromium, it runs AI agents locally on your machine—keeping your data private while giving you superpowers for browsing.

**Open source. Privacy-first. The browser for the AI era.**

## Three Ways to Use AI

<Columns cols={3}>
  <Card title="Chat Mode" icon="message">
    Ask questions about the page you're viewing. Summarize articles, extract data, translate content—without leaving the page.
  </Card>
  <Card title="Agent Mode" icon="robot">
    Describe a task and watch AI execute it: clicking, typing, navigating. Automate form filling, data extraction, and multi-step workflows.
  </Card>
  <Card title="Graph Mode" icon="sitemap">
    Build reliable, repeatable workflows as visual graphs. Edit steps, save, and run anytime. *Coming soon.*
  </Card>
</Columns>

## Quick Start

<Steps>
  <Step title="Download BrowserOS">
    Get the latest release from [browseros.com](https://browseros.com) for macOS, Windows, or Linux.
  </Step>
  <Step title="Import from Chrome">
    Navigate to `chrome://settings/importData` and import your bookmarks, passwords, and settings in one click.
  </Step>
  <Step title="Configure your AI">
    BrowserOS includes a default AI model with limited daily usage for testing. For the full experience, bring your own API keys.

    <Card title="Set up your LLM provider" icon="key" href="/features/bring-your-own-llm">
      Configure Gemini, Claude, OpenAI, or run models locally with Ollama
    </Card>
  </Step>
  <Step title="Try it out">
    Open any webpage and press the **Assistant** button in the toolbar. Ask a question about the page, or describe a task to automate.
  </Step>
</Steps>

## Which AI Model Should I Use?

<Tabs>
  <Tab title="For Chat Mode">
    **Any model works great**, including local LLMs.

    - **Ollama** or **LM Studio** — Free, runs locally, great for quick questions
    - **Gemini Flash** — Fast and affordable
    - **Claude or GPT-4** — Best quality responses

    Chat Mode just needs to understand the page and answer questions—local models handle this well.
  </Tab>
  <Tab title="For Agent Mode">
    **Use Claude Opus 4.5 for best results.**

    | Model | Speed | Quality |
    |-------|-------|---------|
    | Claude Opus 4.5 | Slower | Best |
    | Claude Sonnet 4.5 | Medium | Great |
    | Claude Haiku 4.5 | Fast | Good |
    | Gemini Flash | Fast | Good |

    <Warning>
    Local LLMs are not yet powerful enough for Agent Mode. The model needs strong reasoning to click the right elements and handle multi-step tasks.
    </Warning>
  </Tab>
</Tabs>

## Why BrowserOS?

<Columns cols={2}>
  <Card title="Privacy-First" icon="shield-check">
    Run AI locally with Ollama, or bring your own API keys. Your browsing data never leaves your machine unless you choose.
  </Card>
  <Card title="Open Source" icon="code-branch">
    100% open source under AGPL-3.0. Inspect the code, contribute features, or fork it for your needs.
  </Card>
  <Card title="Works with Your Tools" icon="plug">
    Built-in MCP servers for Gmail, Calendar, Docs, Sheets, Notion. Connect to Claude Code or Gemini CLI for developer workflows.
  </Card>
  <Card title="Chrome Compatible" icon="chrome">
    All your Chrome extensions work. Import bookmarks, passwords, and settings in one click.
  </Card>
</Columns>

## Community

<Columns cols={3}>
  <Card title="Discord" icon="discord" href="https://discord.gg/YKwjt5vuKr">
    Get help and share feedback
  </Card>
  <Card title="GitHub" icon="github" href="https://github.com/BrowserOS-ai/BrowserOS">
    Star us and contribute
  </Card>
  <Card title="Twitter" icon="x-twitter" href="https://x.com/browserOS_ai">
    Follow for updates
  </Card>
</Columns>