Internet Speed Test
Measure your real download and upload speed, ping, jitter and latency under load — live, in your browser, with a result you can share.
You are looking at a shared result. Run your own test to compare.
Test settings
More connections usually measure a fast line more accurately; a single connection is closer to what one download or one video stream will actually get. The data limit caps the download half, which is where nearly all of the data goes — the upload half never sends more than 64 MB.
Throughput
- Download average
- —
- Download peak
- —
- Upload average
- —
- Upload peak
- —
- Connections used
- —
- Data used
- —
Latency
- Idle ping
- —
- Best / worst
- —
- Jitter
- —
- Ping while downloading
- —
- Ping while uploading
- —
- Bufferbloat
- —
Connection
- Your IP address
- —
- Country
- —
- Test server
- —
- Protocol
- —
- Network type
- —
- Tested
- —
What this connection can do
The test sends and receives meaningless random data — nothing from your device is uploaded, and no result is stored on our servers. A full run typically moves a few hundred megabytes, so use Wi-Fi rather than mobile data, or lower the data limit in the settings.
Cómo usar Internet Speed Test
- Close anything else that is downloading or streaming, then press GO.
- Watch the live dial: it measures ping first, then download, then upload.
- Read the full breakdown underneath — peak speeds, jitter, bufferbloat and what the connection can comfortably do.
- Copy the results, copy a share link, or save the result as an image.
Acerca de Internet Speed Test
This measures what your connection is actually doing right now: download speed, upload speed, ping, jitter and latency under load. The dial moves live as the test runs, and when it finishes you get the full breakdown rather than three big numbers — peak versus average speed, best and worst round trip, how much data the test used, which edge server answered, and whether the line stays responsive while it is busy.
The measurement itself is built to be honest, which takes more care than it sounds. The test data is random, because a file of zeros would be compressed in transit and the browser would report a connection hundreds of times faster than it is. The first second and a bit is drawn on the graph but thrown out of the average, because TCP deliberately starts slow and averaging from zero always under-reports — the faster your line, the worse the error. And the upload figure is calculated from bytes the server confirmed receiving, not from the browser's own progress events, which report data handed to the operating system and can run megabytes ahead of what has actually left the building.
The number most speed tests never show you is bufferbloat, and it is often the one that explains the problem you came here about. A router with an oversized queue can hold a perfect 12 ms ping at rest and jump to 800 ms the moment anything downloads — which is exactly why a video call falls apart when someone else in the house starts a backup, on a connection whose headline speed looks fine. This test pings continuously during the transfer and grades the difference, so a fast line with a badly behaved queue is visible instead of hidden.
There is nothing to install and no sign-up. Nothing about you is stored: the upload half is read and discarded, no result is written to any database, and the share link carries the numbers in the URL fragment — the part of a link browsers never send to a server. Your speed depends on Wi-Fi, the device, the time of day and what else is on the network, so for the truest picture of the line itself, test on Ethernet or standing near the router, and run it two or three times.
Preguntas frecuentes
Why is my speed lower than what I pay for?
Usually Wi-Fi rather than the line. A plan sold as 500 Mbps is measured at the socket, and by the time it has crossed a busy 2.4 GHz channel, a wall or two and an older laptop's wireless card, a fraction of that reaching the browser is normal. Test again on an Ethernet cable, or standing next to the router, and compare. Other things on the network count too — one 4K stream or a phone backing up is taking its share while you test.
What is bufferbloat, and why does my grade matter more than my speed?
Bufferbloat is delay caused by your own equipment queuing data instead of dropping it. The symptom is familiar: everything is fine until someone starts a big download, and then calls stutter and pages hang even though the speed test says the connection is fast. This test measures ping while the line is saturated and grades the increase. An A or A+ means the connection stays responsive when busy. A D or F means it does not, and that is worth fixing — usually by turning on Smart Queue Management or SQM in the router — regardless of how high the download number is.
How much data does one test use?
It depends entirely on how fast your connection is, because the test runs for a fixed time rather than a fixed size — a 20 Mbps line moves about 25 MB, a gigabit line can move several hundred. On mobile data, open Test settings and lower the data limit, or shorten the download time. The results panel always tells you exactly how much was used.
Why does the number bounce around while the test runs?
Because it is showing the speed over the last fraction of a second rather than the average so far, which is what makes a Wi-Fi drop-out or a noisy line visible instead of smoothed away. The final figure is the average across the whole measurement window, with the slow-start ramp at the beginning excluded.
What do the connections and time settings do?
Connections is how many transfers run at once. Several at once is the standard way to measure a fast line, because one connection often cannot fill it on its own — but a single connection is closer to what one download or one video stream will really get, so both answers are useful. Longer tests are more accurate on unstable connections and use more data.
Is my result stored, and is the share link private?
Nothing is stored. The uploaded data is random noise that is read and thrown away, and no result is written to any database. The share link puts the numbers in the part of the URL after the # — a fragment, which browsers never send to a server — so opening a shared link tells us nothing about who opened it.
What is a good ping, and what is good jitter?
Under about 30 ms is excellent and under 60 ms is fine for competitive gaming and video calls. Jitter matters as much as ping for anything live: it is how much the round trip varies between one probe and the next, and steady jitter under about 10 ms is what keeps a call clear. A line that alternates between 20 ms and 120 ms is far worse for a call than one that sits steadily at 90 ms, even though the averages look similar.
Which server does it test against?
The nearest edge location of the CDN in front of this site, which the results panel names. That is the right thing to measure for everyday use — it is roughly the distance to a streaming service or a big website — but it does mean a very long-distance connection to one specific server elsewhere in the world can be slower than the figure here.

