ChatBrowserUse() — Optimized LLM for Browser Tasks

Browser-Use’s key innovation is ChatBrowserUse(), a fine-tuned model that’s optimized specifically for browser automation. It completes tasks 3-5x faster than general-purpose models like GPT-4 or Claude while maintaining state-of-the-art accuracy. Pricing is competitive: $0.20/M input tokens, $2.00/M output tokens.

The open-source preview model browser-use/bu-30b-a3b-preview is available for self-hosting, giving users a free option that still outperforms general models on browser tasks.

Persistent Browser CLI

The CLI is a standout feature for developers who want to automate browsers without writing Python:

browser-use open https://example.com    # Navigate to URL
browser-use state                       # See clickable elements with indices
browser-use click 5                     # Click element by index
browser-use type "Hello"                # Type text into focused field
browser-use screenshot page.png         # Capture screenshot
browser-use close                       # Close browser session

The browser stays alive between CLI commands, enabling iterative debugging and rapid prototyping.

Custom Tools System

Agents can be extended with custom Python functions using the @tools.action decorator:

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,
)

This makes Browser-Use extensible for domain-specific automation — anything from database lookups to API calls during browser sessions.

Claude Code Integration

A dedicated Claude Code skill is available, making Browser-Use a first-class tool for Claude-powered development workflows:

mkdir -p ~/.claude/skills/browser-use
curl -o ~/.claude/skills/browser-use/SKILL.md \
  https://raw.githubusercontent.com/browser-use/browser-use/main/skills/browser-use/SKILL.md