Ping test

How this tool works

Enter an IP address or a hostname and our probe server sends a short series of ICMP echo requests to it, the same packets the ping command on your computer uses. Each reply is timed. The result shows how many packets were sent and received, the minimum, average and maximum round-trip time, and the jitter between replies. The test is free and needs no account.

Separately, as soon as the page loads, your browser measures its own latency to our server. It sends a series of very small HTTP requests to a lightweight endpoint and records the time each one takes. Browsers cannot send ICMP, so this figure is an HTTP round trip rather than a true ping, but it reflects the same network path and is the number that matters for how fast this site, and sites hosted near it, feel to you.

The two measurements answer different questions. The probe ping tells you whether a host is up and how far it is from the probe. The browser latency tells you how far you are from our server. Neither one is the latency between you and the host you typed; that path runs through your own connection and can only be measured from your computer, with the commands further down this page.

Hostnames are resolved before pinging. The probe refuses private and reserved addresses, applies a per-visitor rate limit, and caches recent results briefly so repeated runs do not flood a target.

How to read the result

Sent and received. If every packet got a reply, the host is reachable and the path is clean at this moment. If none did, either the host is down, the address is wrong, or something between the probe and the host drops ICMP. Partial replies indicate packet loss somewhere along the way.

Minimum, average and maximum. The minimum is the closest thing to the pure latency of the path: the fastest a packet managed to get there and back. The average is the typical experience. The maximum shows the worst spike during the run. A large gap between minimum and maximum means the path is congested or a router on it is slow to answer.

Jitter. The average difference between consecutive round-trip times. Low jitter means a steady connection. High jitter with a reasonable average is the signature of Wi-Fi interference, a saturated upstream, or a bufferbloated router, and it hurts voice and video calls more than the average latency does.

Packet loss. A percentage of requests that got no reply. 0 percent is the goal. A single lost packet in a short run can be noise. Consistent loss, especially combined with high maximum latency, means a problem worth chasing with a traceroute, which shows at which hop packets start to disappear.

The browser latency block follows the same logic: a minimum close to the average and a small spread means your connection to our server is stable; large swings point at your local network or your ISP rather than at the destination.

What to do about high latency or loss

The host does not reply at all

1. Check the address or hostname for typos and look it up with the DNS lookup to see what it resolves to.
2. Test a port the host should have open with the port checker. A reply on port 443 and silence on ping means ICMP is filtered, not that the host is down.
3. If the host is yours, check whether its firewall drops ICMP echo requests. On Linux look for an iptables or nftables rule on icmp; on Windows, enable the "File and Printer Sharing (Echo Request)" inbound rule in Windows Defender Firewall; on cloud hosts, allow ICMP in the security group.

Latency is higher than expected

1. Compare the minimum with the distance. Light in fiber covers roughly 200 kilometers per millisecond, and a round trip doubles that, so a host 5,000 kilometers away cannot answer in under about 50 milliseconds no matter how good the network is.
2. If the minimum is reasonable but the average and maximum are high, the path is congested. Run a traceroute to find the hop where latency jumps.
3. If the probe shows low latency to a host but your own computer shows high latency to the same host, the slowness is on your side: your Wi-Fi, your router or your ISP's network.

Your own connection shows high jitter or loss

1. Connect by cable instead of Wi-Fi and run the browser latency test again. If the spread collapses, the wireless link is the cause: move closer to the router, change the channel, or use the 5 GHz band.
2. Stop other transfers on the network. A single upload saturating the upstream raises latency for everything else; a router with smart queue management or QoS keeps it in check.
3. Reboot the router and the modem. Aging consumer hardware develops exactly this symptom.
4. Test at different times of day. Loss that appears only in the evening is congestion on the ISP's side, and the ISP needs the measurements as evidence.
5. If you are on a VPN or proxy, repeat the test with it off. Every tunnel adds the detour through its server to your latency; a server on another continent can add hundreds of milliseconds.

Pinging from your own computer

The probe's view is useful for checking whether a server is up and how it performs from the internet. For your own connection, use the ping command.

Windows: ping 203.0.113.45 sends 4 requests by default; ping -t 203.0.113.45 keeps going until you press Ctrl+C and ping -n 20 203.0.113.45 sends 20.

macOS and Linux: ping -c 20 203.0.113.45 sends 20 requests and prints a summary with minimum, average, maximum and standard deviation. On Linux ping -i 0.2 shortens the interval for a quicker read on jitter. For IPv6, ping6 on older systems or ping -6 on newer ones.

Ping your router first, then a host at your ISP, then a well-known public address, then the target. The first hop at which latency jumps or packets go missing is where the problem is.

Ping an IP or ping a hostname

The result is the same once the packets are on the wire, but the two inputs differ in what else they test. Pinging a hostname first resolves it, so a failure can be a DNS problem rather than a network one; the DNS lookup separates the two. A hostname with several addresses, which is common for large sites, is pinged at one of them, and that address may differ from the one your browser connects to. Pinging the address directly removes the ambiguity and is the better choice when you are tracking a specific server.

Ping also tests a different thing than a port check. A host that answers ping is powered on and routed, but says nothing about whether its web server, mail server or game server is running; a host that ignores ping may be serving pages happily. When the question is "is the service up", the port checker answers it; when the question is "is the host reachable and how far away is it", ping does.

Ping and the audience checking a proxy or VPN

For anyone testing a proxy or VPN endpoint before relying on it, two numbers matter: the latency from your own computer to the endpoint, measured locally with the commands above, and the latency from the endpoint to the sites you will use, which the probe approximates when it pings the endpoint from the internet side. Both legs add up in every request you make through the tunnel. An endpoint that is far from you or far from your target sites will feel slow regardless of its advertised bandwidth, and high jitter on either leg is the usual cause of timeouts and dropped sessions.

Limits of this tool

• The probe ping measures the path from our probe server to the host, not from you. Your own latency to the host can only be measured from your computer.
• The browser latency uses HTTP rather than ICMP, so it includes a small amount of protocol overhead and the server's response time. It is a good relative measure, not a laboratory one.
• Many hosts, firewalls and cloud providers drop or rate-limit ICMP. No reply does not mean the host is down.
• Each run sends a small number of packets over a few seconds. Intermittent problems may not show up in one run; repeat the test or use a continuous ping locally.
• Routers often answer ICMP more slowly than they forward traffic, so a single slow hop in isolation is not proof of a problem.
• Hostnames with several addresses are pinged at one of them.
• The tool is rate-limited per visitor and refuses private and reserved address ranges.