Changed inside the engine, not injected
Browz runs its own browser engine. The values a site reads are changed in the browser’s C++ code, so no script is added to the pages you visit and there is nothing in JavaScript for a site to catch out.
Features / Anti-detect
Anti-detect, measured.
Here is every protection in the Browz engine, in plain words, with what public fingerprint test sites say about it and what we measured when we ran other anti-detect browsers through the same tests.
What Browz protects
Fifteen protections, in plain words.
Detectors don’t only read values. They check whether the values agree with each other and with the engine underneath. So Browz aims for one believable device per profile, not the most changes.
One device, not a pile of random values
Each profile is derived from a real device type and a seed. Its platform, screen, fonts, graphics, voices and languages are chosen together, so they agree with each other and stay the same from one session to the next.
Text drawn its own way, without noise
Most tools add random noise to canvas images, and detectors look for exactly that. Browz gives each profile its own small, stable change to how text is drawn. The images stay as clean as a normal Mac’s, but differ between profiles.
Real graphics, chosen labels
WebGL and WebGPU draw with your real GPU, so the pictures they make are genuine. The vendor and renderer names a site reads can be automatic, set by you, or left real. WebGPU can be switched off for a profile.
Audio left native
Audio fingerprints are left as your computer makes them by default, because noisy audio is a known tell. Audio noise is there if you want it.
Fonts and element sizes
The font list matches the device type the profile presents, and the sizes a page measures for text and boxes come from the same fonts, so the two can’t disagree.
Screen, CPU and memory
Each can be automatic, set by you, or real. The default presents your computer’s own, because we measured that detectors score a mismatched CPU, memory or screen as suspicious.
Timezone, language and location follow the exit
The timezone, languages and geolocation of a profile come from its proxy’s exit, so a profile that browses from Frankfurt also keeps Frankfurt time.
WebRTC can’t give away your address
WebRTC is replaced with the exit’s address, or disabled, so a page can’t use a video-call connection to find the real IP of your network.
Port-scan protection
Pages can’t probe the services running on your own computer (localhost), a check some sites use to recognise the same machine behind different accounts. Early connections a page asks the browser to open are covered too.
Automation stays quiet
The webdriver flag and the other traces automation leaves are hidden, so a profile driven by Playwright or Puppeteer reads like a person’s browser.
The network fingerprint is real Chrome
TLS and HTTP/2 come from the engine’s own, unmodified network stack, and match stock Google Chrome exactly, down to the JA4 hash.
User agent and client hints agree
The user agent, the client hints and their brand list match the browser version the profile really runs, and the operating system it presents.
Media devices and voices
The cameras, microphones and speech voices a page can list can be automatic, set by you, or real, so you choose whether a site sees your own hardware.
CSS sees what JavaScript sees
Media queries for the screen agree with the size and pixel density scripts read, a cross-check many detectors run.
Results
What the public test sites say.
What each public fingerprint test site reports for a fresh Browz profile on its default settings.
| Test site | What it checks | Browz result |
|---|---|---|
| BrowserScan | Authenticity score, bot check, and a visitor ID that links visits | Authenticity 100% on 3/3 profiles, bot “No”, three different visitor IDs, each the same after a restart |
| Pixelscan | Whether the fingerprint is consistent, masked, automated or proxied | “Your browser fingerprint is consistent” on 7/7 visits: no masking, no automated behaviour, no proxy detected |
| CreepJS | Lies (values that contradict the engine), headless and stealth scores | 0 lies on 8/8 visits; 0% headless, 0% stealth; a different fingerprint ID per profile, stable after a restart |
| Fingerprint Pro | A commercial fraud detector: visitor ID and a bot-likelihood score | Three separate visitors, each stable after a restart; bot-likelihood score 0.007–0.045 |
| tls.peet.ws | The TLS and HTTP/2 handshake (JA4, Akamai HTTP/2 fingerprint) | Identical to stock Google Chrome: JA4 t13d1517h2_8daaf6152771_cb7bf5808d99, same ciphers, groups and HTTP/2 settings |
| BrowserLeaks WebRTC | Whether WebRTC reveals an address other than the proxy’s | “No leak”; DNS resolved at the proxy’s location |
Detectors update their checks, so we re-run these as the engine is updated.
Head to head
Browz and two leading anti-detect browsers, on the same tests.
Called Browser A and Browser B here. The same detectors and the same number of profiles on each side. These are our own measurements of each product’s default setup on the day; where one of them matched Browz, the table says so.
Browser A (default settings)
| Test | Browz | Browser A | Result |
|---|---|---|---|
| Pixelscan | 3/3 consistent, no masking, no automation | 3/3 consistent, no masking, no automation | Both passed |
| BrowserScan: are profiles told apart? | Three profiles, three visitor IDs, unchanged after a restart | Three profiles, one shared visitor ID | Only Browz passed |
| Fingerprint Pro, secure page | Three separate visitors; bot-likelihood score 0.011–0.048 | The same visitor as stock Chrome; score 0.128; anti-detect flagged | Only Browz passed |
Browser B (the version current on the day)
| Test | Browz | Browser B | Result |
|---|---|---|---|
| Commercial tampering signal | 0 of 4 visits flagged | 4 of 4 visits flagged | Only Browz passed |
| BotD, open-source bot check | Clean 4/4 | Clean 4/4 | Both passed |
| BotD and DeviceAndBrowserInfo, both browsers set to a Windows persona | Clean 4/4 on both | Clean 4/4 on both | Both passed |
| Same, each on its defaults | Clean 4/4 | Flagged as a bot 4/4: its random personas didn’t match the computer | Only Browz passed |
Feature by feature
Browz and four leading anti-detect browsers.
Called Browser A to D here (A and B are the two above). What each browser lets you control, and its default where the docs give one, from each product’s own public documentation, read on 30 Oct 2026. Most of these tools let you change every setting; the difference is in what they do by default and how they do it.
| Browz | Browser A | Browser B | Browser C | Browser D | |
|---|---|---|---|---|---|
| Changes made in the engine, no page scripts | Yes, C++ changes in the engine | Not in its docs | Not in its docs | Yes, per its docs | Not in its docs |
| Canvas | Per-profile text rendering, no noise; noise or block optional | Real (default), noise or disabled | Noise (default) or real | Real advised; noise or off | Noise, off or block |
| WebGL image | Real GPU output; noise optional | Noise (default) or real | Noise (default) or real | Real advised; noise or off | Noise or off |
| WebGL vendor and renderer | Automatic, custom or real | Masked (default) or custom | Real, custom or random match | Real or manual | Mask or real |
| Audio | Real by default; noise optional | Real (default) or noise | Noise (default) or off | Not in its docs | Noise or off |
| Client rects (element sizes) | From the profile’s fonts; noise optional | Not in its docs | Noise (default) or real | Real advised, or noise | Noise or off |
| Fonts | The device type’s set, custom or real | Masked (default) or real | A custom list | Not in its docs | Masked list (advised) or off |
| Screen, CPU and memory | Automatic, custom or real, each | Screen and CPU threads masked, custom or real | Follow the computer, or set values | Real or manual | Set values |
| Timezone and location from the proxy | Yes, or custom | Yes, or custom | Yes (default), or custom | Auto (advised) or manual | Language from the IP; timezone not in its docs |
| WebRTC | Replaced with the exit’s address, disabled or real | Masked (default), custom, real or disabled | Forward, proxy, local or disabled (default) | Off, real, altered or manual | Based on IP, or off |
| Port-scan protection | Yes, early connections included | Yes (default), with an allow list | Yes (default), with an allow list | Yes, a block list you can extend | Not in its docs |
| Media devices | Automatic, custom or real | Real (default), masked or custom | Noise; counts follow the computer or are set | Real or manual | Set counts |
| TLS and HTTP/2 measured against stock Chrome | Identical | Not in its docs | Not in its docs | Not in its docs | Not in its docs |
| Detector results published | Yes, on this page | Not in its docs | Not in its docs | Not in its docs | Not in its docs |
“Not in its docs” means the documentation we read doesn’t mention it, not that the browser lacks it. Browz’s column describes the Browz engine; Installed Chrome profiles have their own controls (see Engines). Prices are compared on the pricing page.
FAQ
About anti-detect.
How does Browz do on public fingerprint test sites?
Browz passes BrowserScan, Pixelscan and CreepJS, and keeps profiles apart on Fingerprint Pro and BrowserScan. Use a good proxy for each profile to keep your accounts apart on the network too.
Why doesn’t Browz add random noise to canvas and audio?
Because detectors look for it. In our tests, noise made Fingerprint Pro’s tampering score jump and Pixelscan report “masking detected”. Browz changes how each profile draws text instead, and leaves audio native. Noise is still there if you want it.
Can I check these results myself?
Yes. Open BrowserScan, Pixelscan or CreepJS in a Browz profile and read the verdict. Use a profile with no extensions and let the page finish loading first.
How do Installed Chrome profiles work?
Installed Chrome runs your own Google Chrome, with controls for CPU cores, timezone, languages and colour profile, and is clean on the open-source BotD detector. The results on this page are for the Browz engine unless they say otherwise.