Running an AI agent against a real website is harder than it sounds. The moment your agent fires up a headless browser, most major sites detect the automation signal and block it before a single task completes. Browser Use just shipped a ground-up rebuild of its cloud browser infrastructure that attacks this problem at the kernel level, not with patches on top of stock Chromium.
What actually shipped
The new stack is built on a custom Chromium fork, a Firecracker fork, and a custom Linux kernel. That's three layers of infrastructure Browser Use now owns and maintains themselves, rather than renting from third parties. It runs on bare metal, costs $0.02 per hour, starts in under a second, and can handle up to 10,000 concurrent browsers.
The new pricing is more than 3x cheaper than the previous offering, which puts it at a price point where browser time becomes a rounding error even across thousands of agent sessions. The infrastructure is live now and applies to all cloud browsers on the Browser Use Cloud.
The stealth problem is worse than most people think
The bottleneck for agents is not their intelligence. On most websites users care about, agents get blocked by a CAPTCHA or an antibot challenge. Browser Use collected 300,000 security check events from production traffic, and the data made it clear: agent intelligence and browser stealth are orthogonal. You need to measure them separately.
Don't miss what's next in AI
Join 300,000+ engineers and researchers who get the signal, not the noise.
- Full access to in-depth AI research breakdowns
- Be the first to know what's trending before it hits mainstream
- Daily curated papers, repos, and industry moves