Public source review
Dated vendor statements, prices, restrictions and unresolved conflicts. No hands-on claim is implied.
BrowserCheck Lab: test methodology
Read the criteria for reviewing official sources and the planned procedures for version consistency, profile isolation and performance tests. A native Chrome reference records 10 reopen cycles; vendor-product tests are pending.
Your environment check
A manual, local observation of this browser. Product identity is self-reported. Export and compare your own results.
Controlled Lab benchmark
Requires installed versions, a fixed device, proxy configuration and repeated runs. The native Chrome reference has 10 reopen cycles. Vendor-product runs remain pending.
Measured reference: native Chrome
Chrome 152.0.7977.83 · macOS 26.5.2 · Apple M5 Pro · 48 GiB · 2026-09-12
Two fresh Chrome profiles on one device, with no account or extensions. The test script supplied no proxy or fingerprint overrides. A visible browser window was controlled through CDP and fully closed between runs. BrowserCheck ran from a local production Worker preview; this is a reference for this setup only.
| Measurement | Observed result |
|---|---|
| Unchanged environment values after reopening A | 10/10 |
| Cookie | Retained in A 10/10 · Absent in B 10/10 |
| localStorage | Retained in A 10/10 · Absent in B 10/10 |
| IndexedDB | Retained in A 10/10 · Absent in B 10/10 |
| Cleanup | All test markers absent; unrelated preference retained |
Same-day short sessions on one device do not establish long-term reliability. The page reports a reduced browser version and generic WebGL renderer; full installed version was recorded separately. RAM, CPU, launch duration, browser updates and other websites were not measured.
AdsPower, GoLogin and Multilogin: awaiting tests
No installed-product runs have been recorded for these three browsers. Their results, versions and measurements remain unpublished until tests can be run with signed-in clients. The Chrome reference cannot be used to rank them.
Research Score: public information criteria
This proposed 100-point rubric measures public information quality and documented capabilities, not anti-detection effectiveness. All current products are unassessed. A score will require a published criterion-by-criterion evidence sheet.
Documentation quality15
Pricing transparency10
Browser / OS coverage15
Automation / API15
Team management10
Data / storage transparency10
Product update activity15
Free trial / accessibility5
Feature differentiation5
Proposed scoring: each criterion receives 0–4 with published anchors and evidence, then contributes weight × rating / 4. Missing or unresolved evidence leaves the overall score unpublished; weights are never silently redistributed.
The installed-product test protocol
Protocol outline v2 · September 12, 2026. Baseline rechecks and Cookie / localStorage / IndexedDB checks are available above. The full protocol below includes tests not yet performed.
Environment consistency
Record User-Agent, Client Hints, platform, architecture, screen, language and time zone. A difference needs investigation; it is not an automatic failure.
Graphics & audio
Repeat Canvas, WebGL, WebGL2 and AudioContext observations under the same configuration. Record unavailable or restricted APIs separately.
Network observations
Run WebRTC and network tests with a documented proxy setup. DNS and TLS require separate instrumentation; a page-only test cannot establish their full behavior.
Profile storage isolation
Write unique disposable markers in cookies, localStorage, sessionStorage, IndexedDB, Cache Storage and Service Workers. Check another profile and clean up afterward.
Lifecycle & recovery
Create, close, reopen, clone, transfer and update profiles. Repeat 10 and 50 launches, log failures, and distinguish intentional resets from lost state.
Performance
Measure launch time, idle CPU and total process-tree RAM at 1, 5 and 10 profiles on one documented device. Use identical pages and report ranges across repeated runs.
What a published Lab record must include
Product and browser versions; operating system and hardware; proxy configuration without credentials; test version and date; repeat count; raw observations; failures; untested fields; and the previous result when a version changes. PASS labels need a defined criterion, not a marketing promise.
Outside the scope
Platform ban rates, account survival, CAPTCHA bypass and universal detection-bypass claims are excluded. Anonymous community submissions are not enabled; no community run counts or verified-brand statistics are being collected.
Sources & commercial relationships
Vendor statements are labelled. No paid placement or affiliate tracking links in this directory. No overall detection score is assigned.
Source reviews & hands-on tests
A source review is not a hands-on test. Research scores remain unassessed until the rubric is completed.
Read our methodology →