Browser Use
@browser_use
Agents that use the Browser | 200k⭐️
5 Following 49.5K Followers
Introducing: Browser Use 1.0.0: CDP is all you need 👀
We deleted 90% of the old library. Predefined state and action space make it worse.
Now the agent just writes Python REPL on raw CDP. Simplicity wins.
> same models, up to 50% better
> 2-5x cheaper
> 2-3x faster
> SOTA web agent
From now on we ship fast. Please pin with ==
Show more
Ran Muse Ember on hardest long horizon browser benchmark 👀
It's EXTREMELY cheap, pareto frontier. Tasks with 100s of browser steps cost CENTS, not USD
> 63.9/100, same as GPT-6 Luna medium
> 3 cents a task, 2.4x cheaper
> best score under 5 cents
congrats @alexandr_wang, cheap models that can actually drive a browser are huge
Show more
BU Ultrafast is absolutely goated for phishing detection.
For less than $0.01 you can detect very robustly whether website is scam or not.
Introducing: Odysseys2 - hardest long horizon benchmark for web agents based on real user tasks 🏗️
We collaborated with @JangLawrenceK @rsalakhu from Carnegie Mellon University
> SOTA agent fully completes only 47/100
> 5 agents complete none
> 8-12 real rubrics per task, 60 min on the live web
Web agents are becoming better than humans at knowledge work.
Show more
GPT-6.1 Terra is insanely good at browser use 🔥
> 78.5 on the Browser Use Benchmark
> basically Sol performance (79.7) at 35% less cost.
Pareto frontier is once again dominated by OpenAI.
Show more
Kimi K3.1 Pro is a beast at browser use 🤯
76.8 on the Browser Use Benchmark. Only GPT-6.1 Sol scores higher.
It beats Opus 5.5, Grok 4.7 and Sonnet 5.5, and it's cheaper than Opus at ~$2/task.
Best Chinese model we've ever tested, by a lot. Next best is DeepSeek V4.1 Flash at 70.6.
Moonshot cooked.
Show more
Macos harness 0.2 is out. Best in class computer use on mac
Shipped MacOS Harness 0.2: massive improvements -it's actually amazing now 👀
Never use your mac again - with any harness.
> virtual cursor
> supports all harnesses
> native integration with Browser Use
> codex computer use level
The agent will use MacOS Harness and then switch to Browser Use when controlling the browser.
Show more
Check it out, we are one of the tools
We’re giving away $1 million to agents.
Introducing Composio Instant: the easiest way for agents to pay for their own tools.
Claim your agent's share:
Show more
Angelina Yang wrote about the story behind Browser Use.
If you want an introduction to what Browser Use is, how it works, and how it differs from traditional automation scripts, have a look!
Link 🔗⬇️
Show more
Literally cheaper than scraping APIs
Browser Use has gotten 100x cheaper this year.
Browser Use has gotten 100x cheaper this year.
Haiku 5.5 on Browser Use Bench v2.1 👀
> same performace as Luna
> 2.8x more expensive
Haiku gets 5x more expensive after 100k context, including cached tokens. BU Bench 2.1 is mostly extremely long-running horizon tasks. It hits that pricing tier a lot.
Show more
.@browser_use put together a toolset for Claude's new computer use and browser tools.
If you already drive a browser with it, this is the on-ramp to wire it into Claude without writing the glue yourself.
Show more
You can now use browser-use with the new anthropic computer SDK.
Computer use and browser use toolsets are now built into the Python and TypeScript SDKs for Claude.
The API tells you what Claude wants to click or type.
Previously, you had to write your own loop and map clicks and keystrokes to commands, but the SDKs now run the loop and send actions to drivers.
Show more
Really cool seeing our on-prem stealth browser powering Gumloop’s Agent Browsers.
This is exactly what we built it for 🫡
Automating stuff where API or MCP don't exist.
Agent Browsers are live on Gumloop today so your agents can reach the work that doesn't have an MCP or API
> Secure credential storage & 1Password integration
> Session replays to see your agents work
> Stealth, proxied sessions
All completely model agnostic
Try Browser today.
Show more
For anyone playing the cat and mouse game, we have the best stealth on the market 👀
The most important technology problem of the next 5 years:
Creating a broadly accepted standard for agents to identify themselves to service providers, so that providers can adjust the way they interface with clients (as compared to human-based traffic).
In the interim (i.e., for the next 6 months), there will be a cat-and-mouse game that agents will play -- to circumvent detection and maintain their product's utility during this growth phase.
However, this will only be a stopgap and it will not be the terminal state of the world.
Show more
We make sure you will never use the browser
