The new GPT-5 performs worse than Opus 4.1 in Stagehand evals in both speed and accuracy. The smaller models are faster, but also still fall short of Opus 4.1.

Every company has a use-case for Browserbase. See what people are building, and how you can supercharge your scraping, automations, and agents with Browserbase.

Slack Operator already exists! Use computer-use models in Stagehand 🤘, and have agents perform tasks on your behalf, All from a simple message in Slack. Source code linked below.

Today, we've been featured on the Forbes Next Billion Dollar Startups 2025 list. We're grateful for this recognition and excited to continue building a product that developers love.

Your onboarding SUCKS, but it doesn't have to. Customers are everything and they deserve a custom experience. Pull information about a sign up and customize their dashboard while they're still onboarding. It's OSS (and using Browserbase).

🚨 Stagehand Evals Leaderboard Update🏆 Gemini is back on top with 2.5 pro in terms of accuracy, and takes the cake when it comes to accuracy/cost. A not so surprising Opus 4.1 is 2nd place in accuracy, but is MUCH more expensive per task. OpenAI's models trail behind, with

You've seen us before, but what does Browserbase actually do?

Stagehand 2.4.3 is live!🤘 • Shadow DOM support • Fix for same-process iframes • Enabling scrolling within iframes • Handle namespaced elements in xpaths • Bump Zod to be compatible with v4 • Patch for new GPT-5 API format Write bulletproof automations, use Stagehand.

Bulletproof your browser automations. We've partnered with Temporal to show you how easy it is to get started using Stagehand with Temporal. Get started using Temporal for durable execution of your Browserbase workflows.

🚨 Stagehand Evals Leaderboard Update🏆 OpenAI's open source models on Groq Inc are very fast and accurate. With an average inference cost of $0.003 per task and 86% overall accuracy on our benchmark, it's the most performant model from OpenAI

Today we're announcing an unlikely partnership. We believe that agents need reliable, responsible web access. That's why we're partnering with Cloudflare in support of Web Bot Auth and Signed Agents, a new standard to allow good bots to authenticate themselves. Details 👇

Excited to announce we've been named to the Madrona' 2025 IA40 List. We're grateful to be recognized with other amazing companies like fal, CrewAI, Vercel, and more.

Stagehand Agent can now use MCP tools.⚒️ Simply pass in the server url, and tell the agent to use the tools. Watch Stagehand use Exa, Supabase, Notion, and Stripe MCPs. 🧵

Only in SF would Legos generate 250k in pipeline. Last week Erika⚡️⚡️ and I hand-delivered custom legos and demos to startups that recently raised. If you want a custom minifigure (and to learn how Browserbase can supercharge your agents and automations), my dms are open!

Password protected site? No problem. Director allows you to take over the Browser to login to any website, then saves the login in contexts. Log in once then stay logged in during future Director sessions, without giving AI your passwords.

Scraping data shouldn’t be a headache. That's why we're partnering with MongoDB to show how easy it is to extract and insert any structure of data into Atlas. Scale your automations and scraping, read more below.

Extract any structure of data you need using Stagehand, Simply pass in a Zod Schema and get exactly what you want from any web page. Try it out on any website!