DNS Benchmark - Compare Public DNS Resolver Speed
Benchmark the major public DNS resolvers - Cloudflare 1.1.1.1, Google 8.8.8.8, Quad9 9.9.9.9, and OpenDNS 208.67.222.222 - with live, server-measured response times. We send real DNS queries to each one, time every query, average them, and rank the resolvers fastest-first. No numbers are fabricated: a resolver that does not answer is shown as "no response", never as a guess.
Fastest resolver for example.com: Quad9 at 10.58 ms average
| # | Resolver | IP | Avg | Min | Max | Success | Status |
|---|---|---|---|---|---|---|---|
| 1 | Quad9 Global anycast | 9.9.9.9 |
10.58 ms | 4.76 ms | 21.89 ms | 4/4 | Responded |
| 2 | Cloudflare Global anycast | 1.1.1.1 |
11.61 ms | 6 ms | 22.46 ms | 4/4 | Responded |
| 3 | OpenDNS Global anycast (Cisco) | 208.67.222.222 |
21.63 ms | 20.83 ms | 22.05 ms | 4/4 | Responded |
| 4 | Google Global anycast | 8.8.8.8 |
24.7 ms | 20.98 ms | 27.68 ms | 4/4 | Responded |
These timings are measured from our server's network location, not from your computer. The resolver that is fastest here may be slower for you, because DNS latency depends on how close the resolver's nearest anycast node is to you. Treat this as a like-for-like comparison of the resolvers' general responsiveness, and test from your own machine before switching.
What is a DNS benchmark?
A DNS benchmark measures how quickly a DNS resolver answers a query. Every time you open a website, your device first asks a resolver to translate the human-readable name (for example example.com) into the IP address your browser actually connects to. That lookup happens before a single byte of the page loads, so a slow resolver adds a visible delay to everything you do online. This tool sends real queries to each major public resolver, times the round trip with a microsecond clock, repeats it several times, and ranks the resolvers by their average response time so you can see which one answers fastest right now.
The numbers on this page are not estimates or cached marketing figures. They are measured live at the moment you run the benchmark, from this server's network, using the net_dns2 library, which can send a query to one specific nameserver and time exactly how long the answer takes to come back.
How does this DNS benchmark actually measure speed?
For each resolver in the list, the tool does the following:
- Creates a resolver client pointed at that one nameserver IP (for example
1.1.1.1for Cloudflare). - Records the time with
microtime(), sends anA-record query for the test domain, and records the time again when the answer arrives. The difference is the measured response time in milliseconds. - Repeats the query a few times. The very first query is treated as a warm-up and dropped from the average, because the first lookup also pays for connection setup and an un-warmed cache. The remaining queries are averaged, and the fastest and slowest are reported too so you can see how consistent each resolver is.
- If a resolver does not answer within the timeout, it is marked no response and excluded from the ranking - it is never assigned a made-up number.
Because every resolver is asked to resolve the same domain, the same number of times, in the same run, the comparison between them is fair. What it measures is the round-trip latency plus the resolver's own processing time, which together are what you feel as "DNS is slow".
Why is the fastest resolver here maybe not the fastest for you?
This is the single most important caveat, so it is worth stating plainly: DNS speed is local. The big public resolvers all use anycast, which means the single IP 1.1.1.1 is announced from hundreds of data centres worldwide, and your query is routed to whichever one is nearest to you on the network. The latency you experience is dominated by the physical and network distance between your connection and that nearest node.
This server has its own location and its own set of nearest nodes, which are almost certainly different from yours. So a resolver that wins by a few milliseconds here might lose for a visitor on another continent, on a different ISP, or behind a different peering arrangement. Use this benchmark to understand the resolvers' general behaviour and to spot a resolver that is clearly struggling, then confirm with a benchmark run from your own machine before you change your settings. A few milliseconds of difference rarely matters in practice; a resolver that is consistently hundreds of milliseconds slower, or that frequently fails to answer, is the one worth avoiding.
What do the numbers mean - average, min, and max?
The average is the headline figure used for ranking: the mean response time across the successful queries (after the warm-up). The min is the best single result, which shows the resolver's potential when everything lines up, and the max is the worst, which hints at jitter or an occasional slow path. A resolver with a low average and a small gap between min and max is both fast and consistent - the ideal. A resolver with a low min but a high max is fast when it is hot but unpredictable, which can produce the occasional frustrating pause. The success column shows how many of the attempted queries actually returned an answer; anything less than a full success rate is a reliability warning even if the timings look good.
Does a faster DNS resolver make my whole internet faster?
It makes the first step of every new connection faster, which is more noticeable than people expect because a modern web page can pull resources from dozens of different domains, each needing its own lookup. Faster DNS shaves a little off the start of each of those. But once a name is resolved, the answer is cached on your device and your network for the lifetime of its TTL, so repeat visits to the same site do not re-query DNS at all. The practical upshot: a faster resolver gives you a snappier feel when browsing to new sites and across many domains, but it does not increase your bandwidth or speed up a large download once the connection is established. If your browsing already feels instant, the DNS resolver is probably not your bottleneck.
Speed is not the only thing that matters: privacy and filtering
Choosing a DNS resolver is not purely a race for the lowest millisecond. Your resolver sees every domain you look up, which is a remarkably complete picture of your browsing, so its logging and privacy policy matter as much as its speed. The resolvers benchmarked here take very different stances:
Cloudflare 1.1.1.1 - privacy-first, no filtering
Markets itself on privacy: it states it does not log the personal data in your queries and purges logs within 24 hours, with periodic third-party audits. The base 1.1.1.1 does no filtering at all; the variants 1.1.1.2 (blocks malware) and 1.1.1.3 (adds an adult-content filter) are opt-in. A good default if your priority is privacy with no content blocking.
Google 8.8.8.8 - fast, ubiquitous, lightly logged
The most widely used public resolver, valued for reliability and global reach. It keeps a temporary sample of full logs for roughly a day or two for debugging, while its permanent logs are anonymised and stripped of your IP address. It performs no content filtering. A safe, fast default, though privacy purists prefer a resolver that logs less.
Quad9 9.9.9.9 - security filtering by default
Run by a Swiss non-profit, Quad9 blocks domains that appear on threat-intelligence lists of malware and phishing, on by default, which can stop a device reaching a known-bad site even before any other protection kicks in. It does not store your IP address. The trade-off is that the security filter occasionally blocks a domain you wanted; for most people that is a feature, not a bug.
OpenDNS 208.67.222.222 - configurable category filtering
Now part of Cisco, OpenDNS pioneered consumer DNS filtering. It blocks phishing and malware out of the box and, with a free account, lets you block whole content categories (adult, gambling, social media), which makes it popular for families and small networks. In exchange it logs more query data than the privacy-first options, feeding Cisco's threat intelligence.
So the right resolver depends on what you value. Want the least logging? Lean Cloudflare. Want built-in protection against malicious domains with no setup? Quad9. Want parental or category controls for a household? OpenDNS. Want a fast, dependable, no-surprises default? Google or Cloudflare. Speed is one input; privacy posture and whether you want filtering are the others.
How do I actually change my DNS resolver?
On most devices you set the resolver in your network adapter or router settings by replacing the automatic (ISP-assigned) DNS servers with two of the IPs above - a primary and a secondary from the same provider, for example 1.1.1.1 and 1.0.0.1 for Cloudflare, or 8.8.8.8 and 8.8.4.4 for Google. Setting it on your router applies it to every device on the network at once; setting it per-device only affects that device. Modern browsers also offer DNS-over-HTTPS (DoH), which sends your DNS queries encrypted to a resolver of your choice and can be enabled independently of the operating system. After changing, re-run this benchmark from your own machine - and use the DNS Leak Test to confirm your queries are really going where you intend and not quietly falling back to your ISP.
DNS benchmark FAQ
Are these DNS speed numbers real or estimated?
They are real and measured live. When you run the benchmark, the tool sends actual DNS queries to each resolver and times the round trip with a microsecond clock, several times per resolver, then averages the results. Nothing is cached from a previous run or copied from a vendor's marketing. A resolver that does not answer is shown as "no response" rather than being given an invented number.
Why might the fastest resolver here be different from what I get at home?
Because DNS latency is mostly about distance. The public resolvers use anycast, so the same IP answers from hundreds of locations worldwide and you reach whichever is nearest to you. This page measures from our server's location, which has different nearest nodes than your connection does. Use this as a general comparison, then benchmark from your own machine before switching.
Which public DNS resolver is the best?
There is no single best - it depends on what you value. Cloudflare leans hardest into privacy and does no filtering by default. Google is fast and reliable everywhere with light logging. Quad9 blocks known-malicious domains by default and does not store your IP. OpenDNS offers configurable category filtering for families. Pick based on whether you want maximum privacy, built-in security, content controls, or just a fast no-fuss default, and let measured speed break a tie.
Why is the first query dropped from the average?
The first query to a resolver in a fresh run pays one-off costs that later queries do not: connection setup and, sometimes, the resolver fetching the answer from upstream rather than its own cache. Including it would unfairly penalise whichever resolver happened to go first. Dropping the warm-up and averaging the remaining queries gives a steadier, more representative figure. We still report the min and max so you can see the full range.
Will changing my DNS resolver speed up downloads or streaming?
No. DNS only handles the lookup that happens before a connection is made. Once a site's name is resolved, the result is cached and the actual data transfer uses your normal bandwidth. A faster resolver makes new connections start a little sooner - noticeable when you browse to many different sites - but it does not increase throughput for a large download or a video stream that is already connected.
Does this benchmark store the domains I test?
No. The domain you enter is used only to run the live queries during that one request and is not stored server-side. The benchmark runs in real time each time you submit, so there is nothing kept between runs.
A resolver shows "no response" - is it down?
Not necessarily. "No response" means that resolver did not answer our queries within the timeout from this server's network during this run. That can happen if a particular resolver's nearest node is unreachable from here, if a network path is congested, or if outbound UDP to that IP is filtered. It is honest about the failure rather than guessing a time. If a resolver fails repeatedly here, try benchmarking it from your own machine before drawing conclusions.
Frequently Asked Questions
Where are these measurements taken from?
Why do big providers like Cloudflare and Google score so close together?
Will changing my DNS server actually make browsing faster?
Why does a resolver show “No response”?
How do I change my DNS server?
How this tool works: This tool runs in your browser and on our server in real time. Depending on the tool, results are computed directly from the input you provide or retrieved from live, authoritative data sources at the moment you run a lookup. We do not sell your data, and your lookups are kept private — any history shown here is stored only on your device.