Snapshot, verified 2026-10-01
Repo:
browser-use/browser-use(formerlygregpr07/browser-use), MIT, ~116.9k GitHub stars (2026-10-01). PyPI latest 0.13.10 (2026-09-04), Python >=3.11,<4.0 (the GitHub releases page shows the same tag 0.13.10). Release notes: Browser Harness 0.1.13, MCP Python SDK 2.1.1, pypdf security upgrade.
Current usage (README, 2026-10-01)
Install: uv add browser-use (or pip install browser-use). Three usage paths: fully hosted Cloud API, a CLI tool, and the open-source Python library.
import asyncio
from browser_use import Agent, ChatOpenAI
from dotenv import load_dotenv
load_dotenv()
async def main():
agent = Agent(
task="Find the number of stars of the browser-use repo",
llm=ChatOpenAI(model='gpt-5.6-luna')
)
history = await agent.run()
print(history.final_result())
asyncio.run(main()) - Models: Browser Use’s own optimised BU2, plus Claude, GPT, Gemini (and local via Ollama per earlier check).
- Custom tools: registered through a
Toolsdecorator (replaces the 2024Controller). - Cloud browsers: stealth mode, CAPTCHA handling, residential proxies; real Chrome profile reuse.
Browser Use Cloud plans (browser-use.com/pricing, 2026-10-01)
Pay-as-you-go credits, no subscription, no expiry. Browsers are billed per hour plus traffic (residential proxy traffic costs more than direct); managed Browser Use Agents at model cost plus a 20% fee; 1-minute browser billing minimum; 10 concurrent sessions at start, scaling to 1,000 with lifetime spend; eligible new signups get free credit; enterprise custom (HIPAA). The page lists per-model token rates, which can differ from vendor list prices. Current prices: see the vendor pricing page https://browser-use.com/pricing
What changed since the Dec 2024 note
- The 2024 quick start (LangChain
ChatOpenAI,Controller,BrowserConfig/BrowserContextConfig, claude-3-5-sonnet, gpt-4o, “any LangChain LLM”) is legacy and has been removed because the current API uses nativebrowser_usechat classes andTools. See the 2024 original in the review backup if needed. - Repo moved org; stars grew to ~117k; hosted cloud and BU2 model added.