Internet Speed Test
Local only — nothing is uploaded unless you save or share it.
An internet speed test that keeps the history — download, upload, ping and jitter, measured properly.
- Press start — it takes about twenty seconds
- Read the download and upload speed, the ping and the jitter
- Save the result to build a history of what your connection actually does
Download
Nothing has been measured yet.
Measured from this browser to a nearby server. On WiFi you are often measuring the WiFi rather than the line — test with a cable before ringing your provider.
Your test history
By time of day
Median download for each hour you have tested. The evening dip is the thing most people are looking for.
Sign in and every test is kept here automatically — one reading tells you nothing, a fortnight of them settles an argument.
Save this tool
Keep Internet Speed Test handy
Use your browser's Share menu, then “Add to Home Screen”.
About this tool
Everybody runs an internet speed test at the moment the internet feels slow, gets a number, and learns nothing — because one reading taken during the problem cannot tell you whether the problem is normal. People check internet speed at exactly the wrong moment, and the useful question is not "how fast is it now" but "is it usually this bad, and at what time of day", and that needs a history rather than a number. This is a download speed test and an upload speed test in one pass, with latency and jitter alongside, measured against Cloudflare's network, which this whole site runs on — so the far end is a server near you rather than one machine somewhere popular. It takes several samples, throws away the first, and reports a high percentile of the rest, because a single transfer measures TCP warming up as much as it measures your line. Then it keeps them. Save a test and it joins a history you can look at later: the same connection at 9am and 9pm, before and after the engineer visited, on WiFi and on a cable. That is the evidence an ISP asks for and almost nobody has, and it is the one thing no chatbot can do for you — it cannot have measured your line last Tuesday. What it cannot tell you, said here rather than implied: this measures this browser to a nearby server over one path at one moment. It is not the theoretical maximum of your line. If you are on WiFi you are very often measuring the WiFi rather than the connection — the single most common reason a paid-for 500 Mbps reads as 80. Test with a cable before you ring anybody, and take a few readings at different times before you conclude anything.
Frequently asked questions
Why is my speed lower than what I pay for?
Usually WiFi. A connection sold as 500 Mbps very often reads as 80 over wireless, especially on 2.4 GHz, through a wall, or on a router a few years old — and the test is honestly reporting what the browser can achieve, which is the WiFi. Plug into the router with a cable and run it again: if the number jumps, the line is fine and the wireless is the problem. Other common causes are other people in the house, a busy evening on a shared line, and an old device whose network card cannot go faster.
How is the speed worked out?
Several transfers rather than one, with the first discarded and a high percentile of the rest reported. That is not tidying the number up — a single short transfer mostly measures TCP's ramp-up, which is why naive tests read low on fast connections, and the first sample also carries DNS and TLS setup. The mean would be dragged down by the same artefacts, so the figure shown is what the line does when it is working.
What is jitter, and why does it matter more than speed?
Jitter is how much the round-trip time varies between one moment and the next. For anything live — a video call, a game, a voice call — it matters more than raw speed: a connection that alternates between 20 ms and 120 ms is unusable for a call while looking perfectly fast on a download test. Under about 30 ms is comfortable; over 50 ms you will hear it. A ping test alone will not show it — you need several samples to see the variation at all, which is why this takes eight.
What is a good speed?
It depends entirely on what you do. About 3 Mbps carries a video call, 10 handles HD streaming and comfortable working, 25 is smooth 4K, and above 100 you are into territory where the speed is rarely what limits anything. Latency and jitter matter more than raw speed for calls and games, and upload matters far more than most people expect if you are on camera.
Why keep a history?
Because one reading is not evidence. "It is slow" is an opinion; twenty timestamped readings showing 40 Mbps every evening on a 500 Mbps line is a case. Save the tests you run and you build that without thinking about it — and you will also find out whether the problem is the time of day, which is the answer surprisingly often.
Does this use a lot of data?
A full test moves roughly 30–50 MB depending on how fast your connection turns out to be, because a fast line needs a larger transfer to measure accurately. On a metered or mobile connection that is worth knowing before you run it several times.
Is anything about my connection stored?
Only what you choose to save, and only to your own account. Nothing is recorded unless you press save, and results are not published, aggregated or sold. The test itself sends random bytes to a server and back — there is nothing in the payload but noise.
Help improve this tool
Found a bug, want a change, or need a tool we don't have?
You have an unsent draft.
Thanks — that's with us. You'll find the reply in your inbox.Go to my messages