Ping / latency check
Measure round-trip time from your browser to three well-known servers.
Click Start test to measure round-trip time to three well-known servers. 10 samples per target.
Measurements are HTTP load time, not ICMP ping — numbers will be
slightly higher than a command-line ping, but relative
differences are accurate. A 5000 ms reading means the request
timed out (the target is unreachable or heavily slow).
About this tool
This measures round-trip time from your browser to three well-known hosts — Google, Cloudflare, and GitHub — taking ten samples each so a single hiccup doesn't skew the picture. Lower is better: latency is the delay before data starts moving, and it's what makes video calls laggy and games feel sluggish even when raw bandwidth is fine. A bar that hits 5000 ms means that request timed out rather than completing slowly.
Because browsers can't send raw ICMP packets — the protocol the command-line ping uses — this times how long a small HTTP image load takes instead. That adds the overhead of a TCP and TLS handshake on top of the pure network delay, so the numbers here run somewhat higher than a terminal ping to the same host. That's expected; what matters isn't the absolute figure but the comparison between the three.
Read it as a quick diagnosis. If all three hosts are slow, the bottleneck is your own connection — your Wi-Fi, your router, or your ISP's first hop. If one host is slow while the others are quick, the problem is that specific service or the route to it, not you. As a rough guide, under ~50 ms feels instant, 50–150 ms is fine for most things, and consistently above ~200 ms is where calls and gaming start to suffer. Everything runs from your browser; no results are uploaded.
How to use this tool
-
Click Start
The test fires 10 samples each to Google, Cloudflare, and GitHub — three of the largest networks on the internet, each on different infrastructure.
-
Wait for the bars to settle
The chart updates live as samples land. Final min/avg/max/median numbers appear after about ten seconds, once all 30 probes complete.
-
Compare the three hosts
If all three are slow, the bottleneck is your local connection. If one is slow and the others are fine, the problem is that specific network or the path between you and it.
Frequently asked questions
Why is this different from command-line <code>ping</code>?
Browsers can't send raw ICMP packets (the kernel-level protocol ping uses), so this tool measures HTTP image load time instead. That includes the TLS handshake, any server-side processing, and image decode — so readings will be noticeably higher than ICMP ping. But relative differences between targets and whether a target is reachable at all are still meaningful.
Why these three sites?
They each sit on a different major CDN / network (Google, Cloudflare, and Azure via GitHub), are globally distributed, serve a tiny favicon reliably, and allow cross-origin image loads. Testing all three gives you a rough picture of internet-wide responsiveness from your browser right now.
What does a 5000 ms bar mean?
That the request timed out — we cap each sample at five seconds so the test doesn't hang. Common causes: the target is genuinely unreachable from your network, a browser extension or firewall is blocking it, or the network is badly congested. Try again, or check the other targets to see if it's specific to one site.
Why are these numbers higher than command-line ping?
Browsers don't have access to ICMP, so the measurement is the time to load a 1×1 image over HTTPS. That includes TLS handshake overhead and connection setup that ping skips. Expect numbers ~20-40 ms higher than a CLI ping to the same host.
What counts as good vs bad latency?
Under 50 ms is excellent — typical of fibre or cable on the same continent. 50-150 ms is normal across an ocean. Above 200 ms is sluggish for interactive work (SSH, video calls). Above 500 ms there's almost always a problem.
Troubleshooting
- All three bars hit 5000 ms
Every request is timing out, which usually means you're offline, stuck behind a captive portal (a hotel or airport login page), or a firewall is blocking the test. Confirm you can load other sites first, then retry.
- The numbers are far higher than command-line ping
This times an HTTP request, which carries TCP and TLS handshake overhead that raw
ICMPping doesn't. The absolute figure will always read higher; use the comparison between the three hosts, not the raw milliseconds.- One host is consistently slower than the others
That points to the route or peering to that specific service, or simply its distance from you — not your connection. If the other two are quick, your link is fine.
- Results jump around between samples
Wi-Fi interference or a busy network makes latency unstable. Test on Ethernet with other devices idle for steady readings; the ten-sample average smooths out the worst of the noise.