I ran PrivacyTests.org’s checks on 7 Linux browsers, including my own. Results, limits and raw data inside.

I make RUUN, a Tor Browser fork with two modes chosen at launch: Tor Network (forgets everything, like Tor Browser) and Direct Internet (normal browsing, history and cookies kept, uBlock Origin built in). It is not an official Tor Project product.

I wanted to know how it compares, so I ran the same tests on 7 browsers on one Linux machine. I ran these tests myself, on my own browser, and I used Claude (an AI assistant) to help write the test tool and the write-up. Please treat this as a claim to check, not a verdict. The tool and the data are public so you can rerun it.

What this is not: it is not an official PrivacyTests.org score, and it is not a ranking. Browsers are in alphabetical order on purpose. The two tables are in the images.

1. Privacy: checks passed out of 116 (PrivacyTests.org pages, middle of three rounds; Tor Browser had two)

– Brave 1.96.59: 86 (blocks trackers, keeps data after a restart)

– Chrome 154: 25

– Firefox 156.0.1: 34

– Mullvad Browser 15.0.23: 102 (blocks trackers, forgets everything on close)

– RUUN Direct mode: 78 (uBlock Origin, but keeps history and cookies on purpose)

– RUUN Tor mode: 76 (through Tor, forgets everything, no tracker blocking on purpose)

– Tor Browser 15.0.23: 76 to 77

What to take from it:

– Mullvad scored highest. RUUN is not the best browser in these tests.

– RUUN Tor gave the same answers as Tor Browser on every check except the noisy tracker ones.

– Tor-based browsers give up points on purpose, because blocking trackers makes you easier to tell apart. A higher score is not the same as safer.

– RUUN Direct went from 55 to 78 when uBlock Origin was added. It fails one test (ECH) because I keep encrypted DNS off on purpose.

2. Speed and memory (fresh profile each time, middle of three rounds, lower is better)

– Brave: 414 MB, starts in 0.6 s (*), 3 idle connections

– Chrome: 467 MB, starts in 0.6 s (*), 5 idle connections

– Firefox: 740 MB, 4.7 s, 12 idle connections

– Mullvad Browser: 700 MB, 4.0 s, 12 idle connections

– RUUN Direct mode: 600 MB, 3.6 s, 5 idle connections

– RUUN Tor mode: 532 MB, 3.6 s, 2 idle connections

– Tor Browser: 594 MB, 3.9 s, 2 idle connections

Idle connections = new outside addresses contacted in 30 seconds at rest. Memory can be compared across all seven. (*) Start time can only be compared inside a family: Brave and Chrome were measured at a different point in startup than the Firefox-based browsers, so 0.6 s does not mean "6 times faster".

The machine and the limits

Ubuntu 24.04, 2 virtual CPUs, 8 GB RAM, no GPU (software rendering), a virtual display, a datacentre internet connection. One machine, one day. Tor Browser has only 2 rounds, because its third round failed four times on network errors, not on test results. The PrivacyTests pages are a fixed list of checks and say nothing about website compatibility, usability, update speed or sandbox strength. A desktop or laptop may give different numbers.

Links

– Full card, with per-category tables, fingerprint values and every limit: https://gitlab.com/varad\_7/ruun/-/blob/main/EVALUATION.md

– The test tool (a Linux port of the PrivacyTests runner) and the cleaned data: https://gitlab.com/varad\_7/linux-browser-privacy-tests

– RUUN itself: https://gitlab.com/varad\_7/ruun/-/releases/v1.0-rc.3

I would like to hear what is wrong with the method, and which browsers or tests I should add.