VPN Leak Tests Compared
dnsleaktest, ipleak, browserleaks, whoer and ours, run side by side: how many DNS names each looks up, whether it reads IPv4 and IPv6 apart, what it does with WebRTC, and whether it calls a leak or leaves that to you. They answer different questions, so this is not a ranking.
| dnsleaktest.com | ipleak.net | browserleaks.com | whoer.net | VPNMeter | |
|---|---|---|---|---|---|
| Who runs it, as the site states | IVPN, a VPN provider | AirVPN, a VPN provider (“powered by”) | Not stated on the pages we read | WHOIX LTD, Cyprus (privacy policy) | This site; it sells no VPN |
| DNS: names looked up per run | 6 in the standard test, 36 in the extended one | 300, one after another | 50: 25 over IPv4 only, 25 over IPv6 only | 5 | 6, all at once |
| DNS: time to a result in our run | 7 s standard, 34 s extended | First resolver at 2 s; all 300 in about 7 minutes | 6 s | Resolvers shown, but the page kept changing with its ads past 90 s, so no time to a result | About 9 s for the whole test in Chrome, 9 to 15 s in Firefox |
| DNS: what you see | Each resolver’s address, hostname, ISP and country | Each resolver’s address, country, ISP and hit count | Each resolver’s address, ISP and location, IPv6 resolvers included | Each resolver’s address, provider and country | The resolvers’ operators and how many; not their addresses |
| DNS: does it call a leak? | No - it lists the resolvers and states the rule | No - it lists them and states the rule | No - it lists them | Yes - by country | Yes - by network: a resolver outside your VPN exit’s network is a leak |
| IPv4 and IPv6 read separately | No - one address | Yes | Yes | No IPv6 check seen | Yes |
| WebRTC | On a separate page; shows the addresses | Lists the addresses from its own STUN server, and says when the browser hid the local one | Yes, and compares the public address with your IPv4 and IPv6 | Yes, and marks a public address that differs | Yes, compared with your IPv4; IPv4 only on this page, the WebRTC test adds IPv6 |
| Other checks on the same page | None | Torrent address (opt-in), HTTP headers, system details | TLS and HTTP fingerprints, headers, whois, Tor | An anonymity score of 17 items, headers, blacklists, ports | Time zone against the exit, and a fix guide for each failed row |
| Verdict | Raw data; you judge | Raw data; you judge | Raw data; a verdict on its WebRTC page | A percentage score | Pass, fail or not measured per row, and a headline |
| Other sites contacted in our run | None | None | One (RIPE, for the address’s registration) | About 290, advertising and analytics | ipify, to read your addresses; Google Analytics and Tag Manager, and Cloudflare’s analytics beacon, on the page |
Checked on October 21, 2026, in Chrome and in Firefox on macOS, on a connection without IPv6; first checked on October 4, 2026. Firefox showed the same as Chrome on every row except time. Sites change; how we checked is below.
What the Differences Mean
- More DNS names, more of the resolver pool. A public resolver answers from many addresses. A test that looks up 300 names, or 50, sees more of them than one that looks up 6. To learn whether your lookups leave the tunnel, one resolver outside it is enough; to see every address involved, more names help.
- Listing or judging. dnsleaktest, ipleak and browserleaks list the resolvers and leave the judgement to you. whoer judges by country. We judge by network: a resolver outside your VPN exit’s network is a leak, so a public resolver you chose yourself, such as Cloudflare’s 1.1.1.1, is listed without comment elsewhere and reported as a leak here - your lookups leave the tunnel to it.
- IPv4 and IPv6 apart. A VPN that carries only IPv4 can leave IPv6 going around it. A test that reads one address sees whichever of the two the browser happened to use for it.
- WebRTC compared, or only listed. browserleaks and our tests compare the public address WebRTC exposes with the one sites see; the others show it and leave the comparison to you.
Where Ours Does Less
- It looks up 6 DNS names, against 36 to 300 elsewhere, so it can see fewer of a public resolver’s addresses.
- It shows the resolvers’ operators, not their addresses.
- Our nameserver has an IPv4 address only, so a resolver is seen by its IPv4 address; browserleaks also sees the IPv6 ones.
- Its IPv4 row reads “Protected” for a VPN or hosting network: it cannot tell a VPN exit from any other datacenter address.
- On the leak test page, WebRTC asks one STUN server and compares IPv4 only; the WebRTC test asks two and compares IPv6 too.
- No torrent check, no HTTP headers and no fingerprinting on the leak test page; the fingerprint tests are separate.
How We Checked
Each site was opened in Chrome 155 and Firefox 156 on macOS, with the browser’s ordinary user agent, and its test started the way a visitor would. Every request the page made was recorded, with what it showed and when; at most three runs per site, and nothing clicked beyond starting the test - no torrent check, no camera or microphone, no sign-up. Who runs a site, and what it says it keeps, is taken from its own pages.
What we could not establish: anything on a connection with IPv6 (ours had none); whoer’s time to a result, since its page keeps changing with ads; who runs browserleaks. Safari and phones were not run.
VPN Leak Tests Compared - FAQ
Which VPN leak test is the most accurate?
They measure different things, so there is no single answer. For a list of every resolver address, a test that looks up many names - ipleak or browserleaks - sees more of them. For a judgement of whether your lookups leave the tunnel, one that compares the resolver with your exit, as ours does, says so directly. For IPv6 next to IPv4, use one that reads them apart. Running two that work differently is a sound check.
Why does VPNMeter report a DNS leak when dnsleaktest does not?
dnsleaktest lists the resolvers that answered and leaves the judgement to you. VPNMeter compares each resolver’s network with your VPN exit’s and reports one outside it as a leak - including a public resolver such as 1.1.1.1 you set yourself, since your lookups then leave the tunnel to it. Both can show the same resolver; only the reading differs.
Are these sites run by VPN companies?
dnsleaktest.com says it is run by IVPN, and ipleak.net says it is powered by AirVPN. whoer.net’s privacy policy names WHOIX LTD in Cyprus. browserleaks.com does not say on the pages we read. VPNMeter sells no VPN and carries no affiliate links.
Why compare tests instead of VPNs?
A leak test result is only as good as what the test measures. Knowing what each one checks - and what it leaves out - tells you which result to trust for which question. For VPNs themselves, our lab publishes what it measured, and the provider pages explain each app’s settings.
🧰 Related Privacy Tools
The tests run in your browser. Where one needs to know whose network an address is on, our server looks it up and keeps no address. Nothing is stored unless you choose to share the result.