How we measure

What is measured

• Turtlecute: the "blocked" count out of the "Total" its page showed on the day (132 on 25 September 2026).
• adblock-tester: its score out of 100.

Both pages belong to other people. We did not write them and we do not control them.

How each run goes

1. A fresh browser profile is created for each blocker. Nothing is carried over from an earlier run.
2. The blocker is loaded unpacked, on its default settings. Onboarding pages that must be confirmed to switch it on (Ghostery's) are confirmed.
3. The script waits 60 seconds so the blocker can download and compile its lists.
4. It opens Turtlecute, waits 30 seconds, and reads the count of blocked hosts from the page text.
5. It opens adblock-tester, waits 35 seconds, and reads the score from the page text.
6. Each blocker was run twice. The two rounds matched for every blocker except AdGuard and Adblock Plus, whose Turtlecute counts differed between rounds; the table shows that range. The raw output of those runs is not published.

The browser is Chromium driven by Playwright, headless, with a 1280 × 900 window. A run with no blocker is included as the baseline. The script is tools/compare_blockers.mjs in our public repository.

The numbers, copied from the home page

Measured on 25 September 2026. Higher is better. Supreme AdBlock was version 5.1.2 in that run; the current release is 5.5.1, and we have not rerun the table for it.

Run it yourself

You need Node.js, Playwright, and each competing blocker unpacked in a folder. From the adblock/ folder of the repository:

PW_ROOT=$(npm root -g) node tools/compare_blockers.mjs --ext "uBlock Origin Lite=/path/to/ubol"

Repeat --ext "Name=/path" for each extra blocker. The script always runs "No blocker" and Supreme AdBlock first. Two options exist: --rounds=N (default 1) and --settle=SECONDS (default 60). It prints the date, the browser, the settings and one line per run.

What the numbers do not show

• Test pages change. Their authors can add or remove hosts and checks at any time. Numbers from another day will differ.
• Defaults only. AdGuard and Adblock Plus may do better with extra lists switched on or after a longer set-up. We report their defaults.
• Lab, not field. A headless browser opening two test pages is not your daily browsing. It does not show broken sites, missed ads on real pages, or how a blocker behaves after months of use.
• Two tests are a narrow view. They do not measure YouTube, cookie banners, memory use or page speed.
• Versions move. The table is a snapshot of 25 September 2026, not of today's releases.
• Competitor versions were not recorded. We tested the releases that were current on 25 September 2026 but did not write down their version numbers, so a rerun with today's releases may give different numbers.

The home page also says "60,000+ rules, 36 lists". That is a count of what is built in, not a test result, and a larger count does not mean better blocking.

Page speed

We publish no speed numbers. The repository has a tool, tools/perf_speedtest.mjs, that measures how much an extension slows a page doing Speedtest-style work (a large page, an animation every frame and six parallel connections). It runs on your own computer over loopback in headless Chromium, so its output compares variants with each other, not with your real connection. We use it to catch regressions before a release.

Found a mistake?

Write to info@carbonstealth.eu and we will correct it. Related: Supreme AdBlock and uBlock Origin and status.