Free tool
IPv6 test for any website
Most IPv6 tests check whether your connection supports IPv6. This one checks whether a target website actually works over it. A page only functions over IPv6 if the document host and every embedded resource — scripts, CDN, fonts, analytics — all resolve over IPv6. Enter a URL below to see the full dependency graph.
Last updated: · Free, no sign-up
Method
What does this IPv6 test check?
The checker requests the page once, then reads every URL the returned HTML asks a browser to fetch: scripts, stylesheets, images and their srcset candidates, iframes, video and audio sources, fonts, web app manifests, preconnect and dns-prefetch hints, and url() references inside CSS. It collapses those to a list of unique hosts and resolves both AAAA and A records for each one over DNS-over-HTTPS.
Resolution deliberately goes over HTTPS rather than the machine running the check, so the result reflects what the target publishes rather than whether our own network has IPv6. The verdict then depends on the whole list, not the domain you typed. If you are validating a proxy route rather than a target, the proxy checker is the tool for that job.
The gap
Why a site can publish an AAAA record and still fail over IPv6
An AAAA record covers one host: the one serving the document. Everything else the page loads is a separate hostname with its own DNS, and any one of those that is IPv4-only cannot be fetched over an IPv6-only route. The request fails or silently falls back, and the page renders without its stylesheet, its analytics, or the script that builds half the DOM.
The scale of that gap is measured. Two independent studies of the top 100,000 sites, run four months apart by different teams, agree that roughly 37% are reachable over IPv6. Only 12.6% are fully functional once subresources are counted. That means about two in three IPv6-reachable sites still depend on something IPv4-only, which is exactly the failure a checker that stops at the AAAA record cannot see.
This is also why rotating pools and static assignments behave differently in practice. If you are planning around IPv6 exit capacity, see rotating residential IPv6 pools for high-rotation workloads, or static datacenter IPv6 when a stable address matters more than rotation.
Evidence
How many websites actually support IPv6?
Every figure below is linked to its primary source and was re-checked on July 25, 2026. Site-side and user-side adoption are not the same number and are shown separately.
Site-side: what the top 100,000 publish
| Measure | Value | Source |
|---|---|---|
| Top 100k web hosts fully reachable over IPv6 | 36,864 (36.9%) | Common Crawl, 16 March 2026 |
| Top 100k hosts publishing no AAAA record at all | 61,924 (61.9%) | Common Crawl, 16 March 2026 |
| Top 100k websites fully IPv6-ready, including every subresource | 12.6% | Thottungal Valapu et al., IMC ’25 |
| Top 100k websites IPv6-partial: main page over IPv6, some resources IPv4-only | 24.8% | Thottungal Valapu et al., IMC ’25 |
| Of sites whose main page loads over IPv6, the share that still pull an IPv4-only resource | About 2 in 3 (24.8 of 37.4 points) | Derived from the two IMC ’25 figures above |
User-side: who can reach IPv6
| Measure | Value | Source |
|---|---|---|
| Users reaching Google over IPv6, most recent daily reading | 47.4% (22 July 2026) | Google IPv6 adoption statistics |
| Same measure, trailing 30-day average | 48.2% | Google IPv6 adoption statistics |
| First day the Google measure touched 50% | 28 March 2026 (50.10%) | Google IPv6 adoption statistics |
| Days at or above 50% in the last 60 | 8 of 60, all weekend peaks | Google IPv6 adoption statistics |
| IPv6-capable end users worldwide, 30-day smoothed | 41.9% (22 July 2026) | APNIC Labs |
The user-side number is often quoted as “IPv6 has passed 50%.” It has not, in any sustained sense. The Google series first touched 50% on 28 March 2026 and reaches it on some weekends, when residential traffic dominates; weekday readings sit in the mid-40s and the trailing 30-day average is 48.2%. APNIC, measuring differently, puts worldwide IPv6 capability at 41.9%.
The spread is the point
National averages differ by more than twenty times, so “is IPv6 viable” has no single answer. An IPv6-only pool is a routine choice for a French or Indian audience and a poor one for a Spanish or Egyptian audience.
| Country | Native IPv6 usage |
|---|---|
| France | 86.4% |
| Germany | 75.9% |
| India | 74.6% |
| Saudi Arabia | 67.9% |
| Malaysia | 64.5% |
| United States | 57.8% |
| United Kingdom | 52.2% |
| Italy | 18.6% |
| Spain | 14.3% |
| Egypt | 4.5% |
Reading the output
What the results mean
Pass
The document host and every embedded host publish AAAA records. Nothing on the page needs an IPv4 route.
An IPv6-only pool can render this target completely.
Partial
The page itself answers over IPv6, but at least one host it depends on is IPv4-only or did not resolve.
Expect missing scripts, styles, images, or silently failed beacons on an IPv6-only pool. Check whether the broken hosts matter to what you are collecting.
Fail
The document host publishes no AAAA record, so the connection cannot be opened over IPv6 at all.
This target needs IPv4 or a dual-stack pool.
Inconclusive
The target returned a challenge page, a block, or a near-empty response, so what we measured is not the page a browser would load.
Treat the host list as unmeasured. This is deliberately not reported as a pass: a challenge page has almost no subresources and would otherwise score perfectly.
A per-host row marked IPv4 only is the one worth acting on: that host resolves to an A record and nothing else, so an IPv6-only route cannot reach it. If the address a target sees matters as much as the family it uses, the proxy checker reports exit IP and target reachability for a route you already have.
Honesty
Limitations of this test
We read the HTML as served. Hosts introduced later by JavaScript are not counted, so the real dependency list may be longer than what is shown. A pass is a floor, not a ceiling.
Sites that block automated requests may return a challenge page instead of their real markup. That is reported as inconclusive rather than folded into a pass, but it does mean some targets cannot be measured this way at all.
An AAAA record proves publication, not health. It shows a host has an IPv6 address; it does not prove every path to it works, that the server answers on it, or that your particular IPv6 range is accepted.
This measures the target, not your own connection. If you are trying to find out whether your browser or proxy is leaking an address family, the IPv6 leak test guide covers that case instead.
The host list is capped at 60 unique hosts per check, and results are cached briefly. Pages that exceed the cap are flagged as truncated in the output.
FAQ
IPv6 test questions
What does this IPv6 test check?
It loads the target page once, extracts every host the page asks the browser to fetch — scripts, stylesheets, images, iframes, fonts, preconnect hints, srcset candidates, and CSS url() references — then resolves AAAA and A records for each unique host. The result is the full dependency graph rather than a single lookup on the domain you typed.
How is this different from other IPv6 tests?
Most IPv6 tests check whether your own connection supports IPv6, and the rest check whether one domain publishes an AAAA record. Neither answers whether a target website works over IPv6, because a page fails if any single embedded host is IPv4-only. This test measures the target, not the visitor, and it measures every host rather than one.
Why can a site publish an AAAA record and still fail over IPv6?
The AAAA record only covers the document host. A modern page also pulls scripts, fonts, analytics, and images from other hosts, and any one of those that is IPv4-only cannot be fetched over an IPv6-only route. Measurements of the top 100,000 websites found 37.4% reachable over IPv6 but only 12.6% fully functional, meaning roughly two in three IPv6-reachable sites still depend on an IPv4-only resource.
What does an inconclusive result mean?
The target served a challenge page, an access block, or a near-empty response instead of its real markup. Because such a page has almost no subresources, scoring it normally would produce a perfect IPv6 result for a page nobody can load. The checker reports inconclusive instead and treats the host list as unmeasured.
Does a pass guarantee my scraper will work over IPv6?
No. A pass means every host the page depends on publishes an AAAA record at the time of the check. It does not prove every path to those hosts is healthy, that the target accepts your IPv6 range, or that hosts loaded later by JavaScript are IPv6-ready. Treat it as a strong prerequisite rather than a guarantee.
How many websites actually support IPv6?
Common Crawl found 36,864 of the top 100,000 web hosts (36.9%) fully reachable over IPv6 in March 2026, with 61.9% publishing no AAAA record. Research presented at IMC 2025 found only 12.6% of the top 100,000 websites fully IPv6-ready once subresources are counted. On the user side, Google measured 47.4% of its traffic arriving over IPv6 on 22 July 2026.
Is this IPv6 test free?
Yes. No account, no sign-up, and no payment. Checks are rate limited per address so the tool stays available and so the public DNS resolvers it depends on are not abused.
Should I buy IPv6 proxies if my target fails this test?
No. If the document host publishes no AAAA record, an IPv6-only pool cannot open the connection at all, and no amount of IPv6 capacity changes that. Use IPv4 or a dual-stack pool for that target. Run this check before buying rather than after.
Know the answer before you buy the pool
If your targets pass this test, IPv6 is usually the cheaper way to buy exit capacity. If they fail, no amount of IPv6 fixes it. Check first, then choose.
See IPv6 proxy plans