Keyboard Latency Test

This keyboard latency test measures how quickly a key press reaches the page after a visual prompt, over five rounds, and shows the average in milliseconds. Because the number includes your own reaction, use it to compare two keyboards on the same computer rather than as an absolute delay figure. The same tool also measures how long you hold each key and how fast your system's auto-repeat fires.

How to run the keyboard latency test

1. Click the large Start box. It changes to Wait… with the hint "Hands on the keys. Do not press yet."
2. Keep a finger resting on any key. After a random pause of 1.2 to 3.5 seconds the box turns amber and reads Press!
3. Press the key as fast as you can. The round's result appears in the list under the box, for example "#1 231 ms", and the next round starts on its own.
4. After five rounds the box shows your average. Best shows your fastest single round.
5. Click the box again to run another set of five, or press Reset to clear everything.

If you press during the wait, the box turns red with Too early and that round is discarded. Click it to start again. The pause is random on purpose, so you cannot anticipate it by counting.

Key presses made with Ctrl or Cmd held are ignored, so browser shortcuts do not accidentally count as a round. Tab and Space are captured so the page does not scroll or lose focus mid-test.

What the numbers mean

Under the box, a second line shows the hold range across all presses, for example "hold 62–118 ms over 14 presses", and the auto-repeat detail, "repeat after 498 ms, every 33 ms". Those two appear as soon as you hold any key for about a second, even before you start a round.

What a browser can and cannot measure

A web page receives a key press at the end of a chain. The switch closes, the keyboard's controller scans its matrix and notices, the USB or Bluetooth link carries the report to the computer at its polling rate, the operating system turns the report into an event, and the browser delivers that event to the page. Only the last step is visible to this test. There is no timestamp for the moment the switch closed, so the page cannot isolate the keyboard's own delay from everything after it.

That is why the test is built around reaction time. Your reaction, the display's refresh and the browser's delivery are roughly constant on one computer. If you run five rounds on keyboard A and five on keyboard B with everything else unchanged, the difference between the two averages is mostly the difference between the keyboards. Treat a gap that repeats over three or four sets as real. A gap smaller than the spread between your own rounds is noise.

The display adds a fixed offset too. The box turns amber on the next screen refresh after the timer fires, which is up to 16.7 ms later on a 60 Hz monitor and up to 6.9 ms on a 144 Hz monitor. That offset is the same for every keyboard tested on that screen, so it cancels out in a comparison.

Where keyboard input delay actually comes from

• Debounce. Mechanical switches bounce for a few milliseconds when they close, and the keyboard's firmware waits for the contact to settle before reporting. Typical debounce windows are around 5 ms, and some firmware lets you shorten it.
• Matrix scan. The controller checks the key matrix in a loop. Faster scanning finds a new press sooner; the difference is usually a millisecond or two.
• Polling rate. A USB keyboard polling at 125 Hz is asked for new data every 8 ms, so a press can wait up to 8 ms before it leaves the keyboard. At 1000 Hz the wait is at most 1 ms. Many gaming keyboards advertise this figure directly.
• Wireless link. Bluetooth keyboards add their own interval and often go into a low-power state after a few seconds idle; the first press after a pause can arrive noticeably late or not at all. A 2.4 GHz dongle usually behaves much closer to a wired connection.
• Operating system and browser. Input processing and the browser's event loop add a small, mostly constant amount. A page doing heavy work can delay delivery, so close other tabs while you test.

Added together, the keyboard-side total is typically somewhere between about 1 and 30 ms depending on the model, which is why it is invisible next to a 230 ms human reaction unless you compare carefully.

How to compare two keyboards fairly

1. Use the same computer, the same browser tab and the same display for both.
2. Plug a wired keyboard into a port on the computer rather than through a hub, and make sure a wireless keyboard is awake by tapping a key before you start.
3. Run at least three sets of five rounds per keyboard and write down each average. Ignore your first set if you were still learning the rhythm.
4. Rest your finger on the same key, pressed to the same depth, each time. Pressing from a height adds travel time that has nothing to do with the keyboard.
5. Compare the averages. If keyboard A is consistently 10 ms or more ahead across sets, that is a real difference. If the sets overlap, the two keyboards are equivalent for practical purposes.

A faster keyboard will not change your typing much, but for rhythm games and competitive shooters a consistent 10 to 20 ms advantage is worth having, and this is the simplest way to confirm a manufacturer's claim without extra hardware.

Is Bluetooth keyboard latency noticeable?

For typing, almost never. For games, sometimes. A Bluetooth keyboard reports at its own interval, and most go into a low-power state after a few seconds without a press; the first key after that pause can arrive late or be dropped while the link wakes up. That shows up in this test as one unusually slow round, or a Too early round that you did not cause, rather than as a uniformly higher average.

To measure it, run the five rounds on the Bluetooth keyboard, then plug in a wired keyboard, or connect the same keyboard by cable if it supports both, and run five more. Many keyboards that offer Bluetooth and a 2.4 GHz dongle are faster on the dongle, so test each mode you plan to use. If the averages differ by less than the spread between your own rounds, the wireless link is not the bottleneck.

Hold time and auto-repeat

Hold time is simply how long the key stays down. It is measured for every key you press on the page, including presses outside a round. The range across many presses shows how consistent your tapping is, and an unusually long hold on one specific key, when you did not hold it, points to a sticky switch. The keyboard test on the home page flags that condition by colour; here you see the raw number.

Auto-repeat is produced by the operating system, not by the keyboard. Hold any key for a second and the system starts sending the same key repeatedly: first after an initial delay, then at a steady interval. The test reports the delay before the first repeat and the interval between repeats, and shows the interval as repeats per second. On most systems the defaults work out to a few hundred milliseconds of delay and roughly 30 repeats per second, but both are adjustable. On Windows they live under the keyboard settings as repeat delay and repeat rate; on macOS they are "Delay until repeat" and "Key repeat rate" in the Keyboard settings. If your repeat rate reads far lower than you expect, that setting is the first thing to check, before suspecting the keyboard.

A key that reports repeats without being held is a different problem. That is chatter, a worn switch bouncing on its own, and it shows up as a red key on the home page test rather than as a repeat figure here.

Related keyboard tests

Latency is one dimension of keyboard performance. The keyboard test on the home page confirms that every key registers at all, and marks stuck or chattering keys. The keyboard ghosting test checks how many keys you can hold at once before the keyboard drops one, which matters for the same games where latency matters. If you want to see exactly which key and code the browser receives for each press, including timing-neutral details such as location and modifiers, use what key am I pressing.