---

Browser Use CLI 3.0 正式发布。 为您使用的编码智能体提供一个能可靠操作的浏览器。

将此指令粘贴到 Claude Code、Codex 等工具中：

Install or upgrade browser-use to the latest stable version with uv using Python 3.12, register the skill from browser-use skill, and connect it to my browser. Follow https://github.com/browser-use/browser-use if setup or connection fails.

🌤️ 想跳过安装配置？使用我们的 云服务 ，获得更快、可扩展且内置隐身保护的浏览器自动化！

1. 将您喜爱的编码智能体（Cursor、Claude Code 等）指向 Agents.md
2. 直接开始提问！

1. 安装 Browser Use（Python >=3.11）：

uv add browser-use

或: pip install browser-use

2. [可选] 从 Browser Use Cloud 获取 API 密钥：

.env

BROWSER_USE_API_KEY=your-key

GOOGLE_API_KEY=your-key

ANTHROPIC_API_KEY=your-key

3. 运行您的第一个智能体：

Python 脚本：

python import asyncio

from browser_use import Agent, BrowserProfile, ChatBrowserUse

async def main(): agent = Agent( task="Find the number of stars of the browser-use repo", llm=ChatBrowserUse(model='openai/gpt-5.5'), # llm=ChatBrowserUse(model='bu-2-0'), # Browser Use 的自研优化模型 # llm=ChatOpenAI(model='gpt-5.5'), # llm=ChatAnthropic(model='claude-opus-4-8'), # Sonnet 也表现很好。 browser_profile=BrowserProfile( headless=False, allowed_domains=["*.github.com"], ), ) history = await agent.run() print(history.final_result())

if name == "main": asyncio.run(main())

查阅 库文档 和 云文档 了解更多！

我们在 100 个真实浏览器任务中对 Browser Use 进行了基准测试。完整基准已开源：browser-use/benchmark 。

使用开源智能体

• 你需要 自定义工具 或深度的代码级集成
• 我们建议搭配使用 云端浏览器，以获得领先的隐身性、代理轮转和扩展能力
• 也可以将开源智能体完全部署在自己机器上

使用 全托管云智能体（推荐）

• 更强大的智能体，适合复杂任务（见上方图表）
• 最快捷的启动与扩展方式
• 最佳隐身保护，提供代理轮转和验证码解决
• 1000+ 集成（Gmail、Slack、Notion 等）
• 持久化文件系统和记忆

任务 = "用我的简历和资料填写这份职位申请表。"

示例代码 ↗

任务 = "将这个商品列表加入我的 Instacart 购物车。"

视频：Instacart 购物演示

示例代码 ↗

任务 = "帮我寻找组装定制 PC 的配件。"

视频：PC 配件选购演示

示例代码 ↗

想要更快开始？ 生成一个立即可运行的模板：

uvx browser-use init --template default

这将创建一个带有可运行示例的 browser_use_default.py 文件。可用模板：

• default - 最简配置，快速上手
• advanced - 全部配置选项及详细注释
• tools - 自定义工具与智能体扩展示例

您也可以指定自定义输出路径：

uvx browser-use init --template default --output my_agent.py

Browser Use CLI 3.0 使您的智能体能够以我们迄今为止最高的准确率在线为您完成工作。它由 Browser Harness 驱动，并应用了我们在 智能体驾驭 和 智能体框架 方面的经验：对最新模型而言，给予它们自由往往比抽象掉复杂性效果更好。我们为您的智能体提供直接、可靠的浏览器操作界面。

browser-use <<'PY' new_tab("https://example.com") print(page_info()) PY

该 CLI 允许您的智能体通过 Python 控制浏览器，并在后台管理浏览器。

对于 Claude Code、Codex 及其他智能体，请将此提示粘贴进去：

Install or upgrade browser-use to the latest stable version with uv using Python 3.12, register the skill from browser-use skill, and connect it to my browser. Follow https://github.com/browser-use/browser-use if setup or connection fails.

在 文档 ↗ 中了解更多集成、托管、自定义工具、MCP 等信息

使用哪个模型最好？

我们专门针对浏览器自动化任务优化了 ChatBrowserUse() 。平均而言，它完成任务的速度是其他模型的 3-5 倍，且准确率达到业界领先水平。

有关定价和其他 LLM 提供商，请参阅我们的 支持模型文档。

我能通过 ChatBrowserUse 使用 Claude / GPT / Gemini 吗？

可以。ChatBrowserUse 接受带有提供商前缀的模型 ID，因此只需一个 BROWSER_USE_API_KEY 即可调用所有模型——无需分别提供 OpenAI/Anthropic/Google 的密钥：

python from browser_use import Agent, ChatBrowserUse

llm = ChatBrowserUse(model='anthropic/claude-sonnet-4-6') # 或 'openai/gpt-5.5', 'google/gemini-3-pro' agent = Agent(task='...', llm=llm)

要获得最佳的速度和成本，我们仍然推荐默认的 bu-* 模型。

我是否应该在开源预览模型中使用 Browser Use 的系统提示？

是的。如果您使用 ChatBrowserUse(model='browser-use/bu-30b-a3b-preview') 配合普通的 Agent(...)，Browser Use 仍会为您发送默认的智能体系统提示。

即使您切换到了开源预览模型，也 不需要 额外添加“Browser Use 系统消息”。只有在您想为特定任务有意调整默认行为时，才使用 extend_system_message 或 override_system_message。

如果您追求最佳的默认速度与准确率，我们仍推荐更新的托管 bu-* 模型。使用开源预览模型时，除 model= 参数外，设置保持不变。

我可以为智能体添加自定义工具吗？

当然！您可以添加自定义工具来扩展智能体的能力：

python from browser_use import Tools

tools = Tools()

@tools.action(description='对该工具功能的描述。') def custom_tool(param: str) -> str: return f"Result: {param}"

agent = Agent( task="您的任务", llm=llm, browser=browser, tools=tools, )

这可以免费使用吗？

是的！Browser-Use 是开源的，可免费使用。您只需选择一个 LLM 提供商（如 OpenAI、Google、ChatBrowserUse，或通过 Ollama 运行本地模型）。

服务条款

本开源库基于 MIT 许可证。关于 Browser Use 的服务与数据政策，请参阅我们的 服务条款 和 隐私政策。

如何处理身份认证？

请查看我们的认证示例：

• 使用真实浏览器配置文件 - 复用您已有的 Chrome 配置文件及其已保存的登录信息
• 如需使用带有收件箱的临时账户，请选择 AgentMail
• 要将您的认证配置文件同步至远程浏览器，请运行 curl -fsSL https://browser-use.com/profile.sh | BROWSER_USE_API_KEY=XXXX sh （将 XXXX 替换为您的 API 密钥）

这些示例展示了如何无缝地维护会话和处理认证。

如何解决验证码？

解决验证码需要更好的浏览器指纹和代理。请使用 Browser Use Cloud，它提供专门设计用于避开检测和验证码挑战的隐身浏览器。

如何投入生产环境？

Chrome 可能消耗大量内存，并行运行多个智能体也较难管理。

对于生产场景，请使用我们的 Browser Use Cloud API，它为您处理：

• 可扩展的浏览器基础设施
• 内存管理
• 代理轮转
• 隐身浏览器指纹
• 高性能并行执行

告诉您的电脑该做什么，它就会去完成。

---

Browser Use CLI 3.0 is here. Give your coding agent a browser it can use reliably.

Paste this into Claude Code, Codex, etc:

Install or upgrade browser-use to the latest stable version with uv using Python 3.12, register the skill from `browser-use skill`, and connect it to my browser. Follow https://github.com/browser-use/browser-use if setup or connection fails.

🌤️ Want to skip the setup? Use our cloud for faster, scalable, stealth-enabled browser automation!

1. Direct your favorite coding agent (Cursor, Claude Code, etc) to Agents.md
2. Prompt away!

1. Install Browser Use (Python >=3.11):

uv add browser-use
# or: pip install browser-use

2. [Optional] Get your API key fromBrowser Use Cloud:

# .env
BROWSER_USE_API_KEY=your-key
# GOOGLE_API_KEY=your-key
# ANTHROPIC_API_KEY=your-key

3. Run your first agent:

Python Script:

import asyncio

from browser_use import Agent, BrowserProfile, ChatBrowserUse

async def main():
    agent = Agent(
        task="Find the number of stars of the browser-use repo",
        llm=ChatBrowserUse(model='openai/gpt-5.5'),
        # llm=ChatBrowserUse(model='bu-2-0'),  # Browser Use's own optimized model
        # llm=ChatOpenAI(model='gpt-5.5'),
        # llm=ChatAnthropic(model='claude-opus-4-8'),  # Sonnet also works well.
        browser_profile=BrowserProfile(
            headless=False,
            allowed_domains=["*.github.com"],
        ),
    )
    history = await agent.run()
    print(history.final_result())

if __name__ == "__main__":
    asyncio.run(main())

Check out the library docs and the cloud docs for more!

We benchmark Browser Use across 100 real-world browser tasks. Full benchmark is open source: browser-use/benchmark.

Use the Open-Source Agent

• You need custom tools or deep code-level integration
• We recommend pairing with our cloud browsers for leading stealth, proxy rotation, and scaling
• Or self-host the open-source agent fully on your own machines

Use theFully-Hosted Cloud Agent (recommended)

• Much more powerful agent for complex tasks (see plot above)
• Easiest way to start and scale
• Best stealth with proxy rotation and captcha solving
• 1000+ integrations (Gmail, Slack, Notion, and more)
• Persistent filesystem and memory

Task = "Fill in this job application with my resume and information."

Example code ↗

Task = "Put this list of items into my instacart."

grocery-use-large.mp4

Example code ↗

Task = "Help me find parts for a custom PC."

pc-use-large.mp4

Example code ↗

Want to get started even faster? Generate a ready-to-run template:

uvx browser-use init --template default

This creates a browser_use_default.py file with a working example. Available templates:

• default - Minimal setup to get started quickly
• advanced - All configuration options with detailed comments
• tools - Examples of custom tools and extending the agent

You can also specify a custom output path:

uvx browser-use init --template default --output my_agent.py

Browser Use CLI 3.0 lets your agents do work for you online with our highest accuracy yet. It is powered by Browser Harness, and it applies what we learned about agent harnesses and agent frameworks: the latest models do best when you give them freedom, rather than abstracting away complexity. We provide your agents with a direct, dependable surface for acting in the browser.

browser-use <<'PY'
new_tab("https://example.com")
print(page_info())
PY

The CLI allows your agent to control the browser via Python, and it manages the browser in the background.

For Claude Code, Codex, and other agents, paste this prompt into your agent:

Install or upgrade browser-use to the latest stable version with uv using Python 3.12, register the skill from `browser-use skill`, and connect it to my browser. Follow https://github.com/browser-use/browser-use if setup or connection fails.

Integrations, hosting, custom tools, MCP, and more on our Docs ↗

What's the best model to use?

We optimized ChatBrowserUse() specifically for browser automation tasks. On avg it completes tasks 3-5x faster than other models with SOTA accuracy.

For pricing and other LLM providers, see our supported models documentation.

Can I use Claude / GPT / Gemini through ChatBrowserUse?

Yes. ChatBrowserUse accepts provider-prefixed model ids, so a single BROWSER_USE_API_KEY reaches all of them — no separate OpenAI/Anthropic/Google keys required:

from browser_use import Agent, ChatBrowserUse

llm = ChatBrowserUse(model='anthropic/claude-sonnet-4-6')  # or 'openai/gpt-5.5', 'google/gemini-3-pro'
agent = Agent(task='...', llm=llm)

For the best speed and cost we still recommend the default bu-* models.

Should I use the Browser Use system prompt with the open-source preview model?

Yes. If you use ChatBrowserUse(model='browser-use/bu-30b-a3b-preview') with a normal Agent(...), Browser Use still sends its default agent system prompt for you.

You do not need to add a separate custom "Browser Use system message" just because you switched to the open-source preview model. Only use extend_system_message or override_system_message when you intentionally want to customize the default behavior for your task.

If you want the best default speed/accuracy, we still recommend the newer hosted bu-* models. If you want the open-source preview model, the setup stays the same apart from the model= value.

Can I use custom tools with the agent?

Yes! You can add custom tools to extend the agent's capabilities:

from browser_use import Tools

tools = Tools()

@tools.action(description='Description of what this tool does.')
def custom_tool(param: str) -> str:
    return f"Result: {param}"

agent = Agent(
    task="Your task",
    llm=llm,
    browser=browser,
    tools=tools,
)

Can I use this for free?

Yes! Browser-Use is open source and free to use. You only need to choose an LLM provider (like OpenAI, Google, ChatBrowserUse, or run local models with Ollama).

Terms of Service

This open-source library is licensed under the MIT License. For Browser Use services & data policy, see our Terms of Service and Privacy Policy.

How do I handle authentication?

Check out our authentication examples:

• Using real browser profiles - Reuse your existing Chrome profile with saved logins
• If you want to use temporary accounts with inbox, choose AgentMail
• To sync your auth profile with the remote browser, run curl -fsSL https://browser-use.com/profile.sh | BROWSER_USE_API_KEY=XXXX sh (replace XXXX with your API key)

These examples show how to maintain sessions and handle authentication seamlessly.

How do I solve CAPTCHAs?

For CAPTCHA handling, you need better browser fingerprinting and proxies. Use Browser Use Cloud which provides stealth browsers designed to avoid detection and CAPTCHA challenges.

How do I go into production?

Chrome can consume a lot of memory, and running many agents in parallel can be tricky to manage.

For production use cases, use our Browser Use Cloud API which handles:

• Scalable browser infrastructure
• Memory management
• Proxy rotation
• Stealth browser fingerprinting
• High-performance parallel execution

Tell your computer what to do, and it gets it done.

Made with ❤️ in Zurich and San Francisco