Latency Test
Measure round-trip time to several major public endpoints directly from your browser, a quick, no-install alternative to a command-line ping.
Tool guide
What this tool does
Latency Test measures round-trip time from your browser to several major public endpoints (Google, Cloudflare, Amazon, Microsoft, GitHub), timing each with performance.now() across 3 samples and keeping the fastest, since a browser can’t send raw ICMP ping packets the way a command-line tool can.
A browser has no access to raw ICMP ping packets (that requires OS-level network privileges a web page doesn’t have), but it can still measure a meaningful round-trip time by timing how long an HTTP request takes to complete.
This tool requests several major public endpoints (Google, Cloudflare, Amazon, Microsoft, GitHub) in no-cors mode, which lets the request complete and be timed even without CORS headers (though the response body itself can’t be read), taking 3 samples per endpoint and keeping the fastest to reduce noise from one-off network hiccups.
How to use Latency Test
- Press Run latency test.
- Wait while each endpoint is tested (3 samples each).
- Review the round-trip time results, colour-coded by rough quality.
- Run it again later to see whether results are consistent or tied to network congestion.
Tips for better results
- These times measure DNS lookup, TCP/TLS handshake and server response together, not raw network ping alone, so they’ll typically read somewhat higher than a dedicated ICMP ping tool’s result for the same destination.
- Results naturally vary by your geographic location relative to each provider’s nearest server/CDN edge, there’s no universal "good" number, only a comparison against your own typical baseline.
- Under 50ms is generally excellent (great for gaming/video calls), 50-150ms is generally fine for most everyday use, and above 150ms can start to feel noticeably laggy for latency-sensitive activities.
- Running the test again at a different time of day can reveal whether high latency is a consistent issue or tied to peak-usage network congestion.
Common use cases
- Quickly checking your connection’s responsiveness without installing a dedicated network tool.
- Comparing latency to several major providers to spot an unusually slow route to one specific service.
- Diagnosing a "laggy" feeling connection by checking actual measured round-trip times.
- Establishing a baseline latency reading to compare against later if performance seems to degrade.
Common mistakes to avoid
- Comparing this directly against a command-line ping result and expecting an exact match; this measures full HTTP request time (DNS + TCP/TLS + response), not raw ICMP ping, so it’s a different, typically slightly higher, measurement.
- Treating a single test run as a definitive result; network conditions fluctuate, running the test a few times (especially at different times of day) gives a more reliable picture.
- Assuming one "good" universal latency number applies everywhere; acceptable latency is relative to your location and the specific activity (gaming needs much lower latency than general browsing).
Frequently asked questions
Why can’t this do a real ICMP ping?
Browsers deliberately don’t have access to raw network sockets or ICMP packets for security reasons; this instead times a full HTTP request, which is the closest equivalent measurement available from within a web page.
Why does it test each endpoint 3 times?
To reduce noise from one-off network hiccups or temporary congestion; the fastest of the 3 samples is kept as a more representative result.
Is this the same as a ping command?
No, a command-line ping measures raw ICMP round-trip time alone; this measures a full HTTP request (DNS lookup + TCP/TLS handshake + response), so results will typically read somewhat higher than a dedicated ping tool for the same destination.
What’s a "good" latency number?
Under 50ms is generally excellent, 50-150ms is fine for most everyday browsing, and above 150ms can start to feel noticeably laggy for latency-sensitive activities like gaming or video calls; there’s no single universal threshold since it depends heavily on your location and use case.
Feedback & suggestions
Help us improve Latency Test. We read every message.