Summary of an article published on The Web Scraping Club on 2026-09-06 (free to read). Read the full article
Chrome 152 ships the CPU Performance API. Through navigator.cpuPerformance, a site can read a small integer that places the device in a performance tier, without seeing the processor model, clock speed or core count. It was built so web apps can adapt to the hardware, but it is also a new hardware signal that anti-bot systems can read. So I wanted to know whether it is a real problem for scraping or just another piece of the fingerprinting puzzle.
I cover how the API works and why Chrome introduced it. Then I compare it with the hardware signals already available: WorkerNavigator and hardwareConcurrency, the Device Memory API, WebGPU and the Summarizer API. After that I look at how a scraper can spoof the property, at two different levels, and which stealth browser tools already handle it. The last part is about what anti-bot systems really check, which is whether all the hardware signals tell the same story.
What’s inside
- What the tier values mean, including the special value for a device Chrome can’t classify, and why the tier is static and not tied to current load.
- How the API relates to the Compute Pressure API: one tells you how powerful the device is, the other how busy the CPU is right now.
- Why spoofing at the JavaScript level and patching Chromium itself differ in how detectable they are, and why most scraping teams can’t maintain a custom build.
- Whether switching to Firefox avoids the signal, plus the Chrome version and privacy questions in the FAQ.
In the full article: The code snippets for each API, the DevTools screenshots of their output, the list of stealth browser tools by category and the FAQ on Chrome versions, privacy and Firefox.