Tool
DNS speed test: compare public DoH resolvers from your browser
Run the DNS speed test
The test starts only when you press the button. Your browser then sends a lookup for example.com to each resolver below: one cold request and five warm ones, one resolver at a time. Verified resolvers return an answer the page can read and check; timing-only ones are timed without reading it. Results stay in your browser.
| Resolver | Cold | Warm median | Check |
|---|---|---|---|
| Cloudflare | – | – | answer checked |
| Google Public DNS | – | – | answer checked |
| Quad9 | – | – | timing only |
| Control D | – | – | answer checked |
| DNS.SB | – | – | answer checked |
| CleanBrowsing | – | – | answer checked |
| AdGuard DNS | – | – | timing only |
| NextDNS | – | – | timing only |
| OpenDNS | – | – | timing only |
| CIRA Canadian Shield | – | – | timing only |
| Wikimedia DNS | – | – | timing only |
| LibreDNS | – | – | timing only |
| AliDNS | – | – | timing only |
| Digitale Gesellschaft | – | – | timing only |
| Applied Privacy | – | – | timing only |
| Njalla | – | – | timing only |
| SWITCH | – | – | timing only |
What this DNS speed test measures
This DNS speed test sends real lookups from your browser to seventeen public DNS-over-HTTPS resolvers and times each answer. It asks every resolver for the address of example.com using the wire format defined in RFC 8484, the same request a DoH client makes. Results reflect your connection, your location and the moment you run it.
Resolvers are tested one after another, so they do not compete for bandwidth and distort each other’s numbers. Each one gets a single cold request followed by five warm requests over the same connection. The table then shows the cold time and the median of the warm times, which ignores a single slow outlier.
Cold and warm timings in the DNS speed test
The cold and warm figures answer two different questions. The cold time includes looking up the resolver’s host name, opening a connection and completing the TLS handshake, so it shows what the very first lookup after a quiet period costs. The warm median shows how quickly the resolver answers once the connection is already open.
Browsers and DoH clients keep connections open for minutes at a time, so in everyday use the warm number is the one you feel. A resolver with a high cold time but a low warm median is still a good choice. One that is slow on both counts is likely far away from you on the network or briefly overloaded.
Verified answers and timing only
Browsers protect users by blocking a page from reading responses from other sites unless those sites allow it. Five resolvers in the list allow it, so the test reads their reply and confirms it is a valid answer for example.com. The other twelve answer normally, but the browser hides the content, so only the timing can be recorded.
Timing-only results are still useful for comparing speed, but a fast reply there could in theory be an error page. That is why the table ranks checked answers first and labels the rest clearly. Untick the timing-only option to test the five verified resolvers alone, which gives a shorter run and results you can trust completely.
Reading the DNS speed test table
Each row shows one resolver with its cold time, its warm median and whether the answer was checked. Rows with no answer usually mean the resolver blocked the request, a firewall interfered, or the service was briefly down; rerun the DNS speed test before writing a resolver off, since one failed attempt says little on its own.
A gap of a few milliseconds between neighbouring rows is noise, not a ranking. Look instead for clusters: a group of resolvers under thirty milliseconds warm is likely served from a nearby data centre, while values above a hundred milliseconds point to a distant server. Anycast networks such as Cloudflare and Quad9 usually land in the fast cluster almost everywhere.
Finding the fastest dns resolver for you
Run the test two or three times at different hours before drawing a conclusion, because a single run can catch a resolver during a brief slowdown. The fastest dns resolver from your home may not be the fastest from your office or phone network, so repeat it on each network where the choice matters to you.
Speed is rarely the deciding factor. Differences of ten or twenty milliseconds disappear behind the local cache that browsers and dnscrypt-proxy keep. Once you have a few fast candidates, compare their logging policies, DNSSEC validation and filtering in the shortlist of best DNS servers, or look them up in the resolver directory to see what their stamps declare.
Privacy of the DNS speed test
Nothing happens until you press the start button. The page then contacts each resolver directly from your browser, which means each operator sees a lookup for example.com from your address, exactly as it would from any DoH client. No result is stored on this site, and the copy button only puts the table on your own clipboard.
If you would rather not reveal your address to all seventeen operators, untick the timing-only option or run the test from a network you do not mind exposing. The lookup itself reveals nothing personal, since example.com is a reserved domain kept for documentation and testing, and every resolver receives the identical request.
Frequently asked questions
Why does the cold time differ so much from the warm time?
The cold request has to resolve the resolver's own host name, open a TCP connection and complete a TLS handshake before the query is sent. Warm requests reuse that connection, so they show how quickly the resolver itself answers. Day to day, a browser or operating system keeps connections open, so warm times matter most.
What does timing only mean in the results?
Some resolvers do not allow web pages from other sites to read their answers. The browser can still time the request, but it cannot confirm that a valid answer came back. Those rows are labelled timing only; the answer checked rows prove the resolver returned a proper DNS response for example.com.
Is the fastest dns resolver in this test the best choice?
Not on its own. Speed varies by location, time of day and network, and a few milliseconds rarely matter for browsing. Logging policy, DNSSEC validation and filtering usually matter more. Use the test to rule out slow options, then choose among the fast ones using the shortlist of DNS servers.
Does this DNS speed test measure my system resolver?
No. A web page cannot see or query the resolver your operating system uses. The test measures DNS-over-HTTPS lookups made directly from your browser to each public resolver, which is the path a browser with secure DNS enabled or a DoH client such as dnscrypt-proxy would take. Your system resolver may be a different service entirely.