browser hardware tests
This online keyboard tester shows which keys register, and which never arrive.
Click the board, then press each key. A key lights up the moment its code arrives.
Click the board to begin, then press each key. A key lights up once it registers.
This test reads physical keys. A phone's on-screen keyboard reports no key code, so open this page on a computer to run it. The drawing below is a diagram rather than a readout: it scales to fit the width, so on a narrow screen the caps are too small to read, and the list under it names each key as it registers.
Held now: 0. Most at once: 0.
Hold as many keys as you can at once. The count keeps the most that arrived together.
Nothing has narrowed the layout down yet, so the board is drawn as ANSI 101. Press a key only your keyboard carries and it redraws. Nothing reports the size, so pick yours above. This browser cannot read which character each key types until you press it, so pick your language above or press the keys.
No keys recorded yet. A key you press is named here, so this test reads without the drawing.
The dashed Fn key never lights up. It is resolved inside the keyboard itself and never reaches the computer as a key event, so no page can see it, including this one. The dashed Eject key is not counted. It is an optional media key that browsers on a Mac are not known to report; if yours does, it is listed under the board. The dashed power key is not counted. It belongs to the system, and browsers are not known to report it; if yours does, it is listed under the board.
The latest presses, newest first, as the browser reported them. Held for is how long the key was down: it measures your finger, not how fast the key is. Repeat appears when you hold a key long enough, and it is your system's key repeat setting, not the keyboard.
| Key | Code | Typed | Held for | Repeat |
|---|
Windows and macOS keep some keys for themselves: the Windows or Command key, Alt+Tab. This browser can borrow most of them for the test, in full screen, and then Escape becomes a key under test like any other. Press Stop to leave, or hold Escape for two seconds, which the browser handles even if this page stops answering.
This keyboard test shows which keys reach your browser. Click the board and press each key: a key lights up the moment its code arrives, so a key that stays gray never registered. The board redraws to match your keyboard's layout, and the layout and size selectors correct it when your keys do not settle the question.
A gray key means the browser never received a code for it. That has four common causes: the key itself, the cable or the wireless link, the keyboard's own firmware, or your operating system taking the key first, which is what happens with Print Screen and the Windows key.
This test separates one thing from the other three: it knows whether the code arrived. Narrowing it further takes a second keyboard, or the same keyboard on a second computer.
The test supports all six physical layouts the W3C key code specification defines: ANSI 101, the alternate American 101, ISO 102, Korean 103, Brazilian ABNT 104 and Japanese JIS 106. The test starts on ANSI and redraws when you press a key only another layout carries.
Some key codes exist on certain boards and not others. IntlBackslash sits beside a shortened left Shift on ISO boards such as the UK one, and the drawing switches to ISO when the key beside the left Shift answers. IntlRo appears on Brazilian and Japanese boards, and the input-mode keys beside the spacebar belong to Japanese and Korean ones.
No single key settles it, so the test reads them together. IntlYen sits on a Japanese board and also on an alternate American one, while IntlYen with IntlRo can only be Japanese. When the keys you have pressed fit more than one physical layout, the note under the board says so instead of guessing, and the selector lets you choose by hand.
A single key that types twice from one press is key chatter: the contacts inside its switch bounce, and the keyboard reports a second press you never made. If most keys do it and this test flags no second press, check your system's key repeat delay before you suspect the switches.
To run a double typing test, press the key that types twice several times: the test measures the gap between releasing it and it firing again. Under 50 milliseconds there is no finger fast enough, so the number itself is the proof. A healthy switch settles in about 5 milliseconds, and the fastest double tap a hand makes takes on the order of 100.
One case is different: a key that fires again while two or more other keys are held. A keyboard at the limit of what it reports together can drop a key and send it again, and that looks the same. The test says so when it happens, and pressing that key on its own tells the two apart.
Turn on Ghosting mode, above the board, and hold as many keys as you can. The count keeps the most your browser received at the same moment and names them. Pick a combination such as W, A, S, D, Shift and Space, and the test names each key that never arrived while the others were held.
That count describes your keyboard, cable and computer together. A web page cannot read the keyboard's own rollover rating, and a key the keyboard blocked looks exactly like a key nobody pressed.
Yes. On a Mac the board becomes an Apple extended keyboard: command, option and control named the way Apple prints them, F13 to F19 across the top, fn where a PC has Insert, and clear and equals on the numpad. On Windows the key beside Ctrl reads Win, and on Linux it reads Super. The System selector under the board switches between them if the guess is wrong.
The order changes with it. A PC puts Alt beside the spacebar and the Windows key one place out; an Apple keyboard puts Command beside the spacebar and Option one place out. At every other size the drawing keeps its shape and takes the Apple order, so the key that lights is the one under your finger.
While Command is held on a Mac, no browser reports other keys being released, so they stay lit until you let go of Command.
Pick Laptop in the size selector, then press each key. The drawing matches a laptop's half-height function row and arrows, so a gray key is one your browser never received.
If the top row types nothing, hold Fn and press it again: a laptop can put F1 to F12 on its Fn layer. To tell the laptop's keyboard from the computer, plug in a second keyboard and press the same keys.
Yes. Pick 60%, 75% or TKL (tenkeyless: a full-size board without the numpad) in the Keyboard size selector, and the drawing shows only the keys that size has.
Nothing tells a web page how big your keyboard is. No interface reports the model, the maker or the key count. One interface can read a hardware identifier, but it needs you to pick your keyboard from a device chooser first, it only exists in one browser, and most keyboards expose nothing for it to read.
Working it out from what stays gray does not hold either. A compact keyboard does not lose its keys. It moves them onto an Fn layer, so F1 on a 65% still reports F1 when you hold Fn and press 1.
Yes, and it is measured: Chromium, Firefox and Safari all report the key code, which is what this test reads and what the physical layout detection uses. One combination has a known fault: Safari on a Mac with a European keyboard, covered below.
The caps start on US characters and become yours as you press them: the key event carries the character your keyboard produced, so a ç key relabels itself ç the moment it registers. That works in Firefox and Safari as well. Chromium can go further and relabel the whole board before you start, because it can ask the system directly; Firefox and Safari never added that interface, so there the Language selector under the board labels it from a table instead.
This test's engine last changed on . The page says only what the engine can do, and the two change together.