Keyboard test online
See whether any key stopped responding and how many your keyboard can handle at the same time.
Ready
KEYSSimultaneous keys (rollover)
—
Tested
0/75
While the test is stopped, the keyboard works normally on the page.
Also try
How the test works
Two questions on one screen. The first is whether every key responds: you press it, it lights up on the drawing. The second is how many keys your keyboard accepts at the same time — what the industry calls rollover, and what in a game is the difference between strafing diagonally while crouched and shooting, and your character simply stopping mid-action.
How the test works
- The page listens for keys by physical position, not by letter. That makes the test work the same on ANSI, ISO and other layouts, and it does not get confused when you hold Shift.
- Every key you press lights up bright while held and stays marked afterwards, so you can keep track of what you have already checked.
- Rollover is measured by counting how many keys are pressed at the same time and keeping the highest number that appeared. Hold several together to measure it.
- Nothing is sent to any server. The reading happens inside the browser and disappears when you close the tab.
- While the test is stopped, the keyboard works normally on the page — system shortcuts are never intercepted.
Anti-ghosting and rollover, no fluff
- Ghosting is when you press three keys and the keyboard sends a fourth you never touched, or drops one of the three. It happens because membrane keyboards share traces in the internal matrix.
- Rollover is how many keys the keyboard can report at the same time. "2KRO" handles two, "6KRO" handles six, "NKRO" handles all of them.
- A common membrane keyboard usually caps at 2 or 3 simultaneous keys — enough to break the most basic combo in a game: moving forward, sideways, crouching and shooting.
- A decent mechanical keyboard delivers 6 keys or more. The ones advertising "anti-ghosting" usually guarantee it only on some keys, typically the WASD area.
- The number this test shows comes from the hardware, not the browser: if you got 3, the limit is the keyboard. Worth testing before blaming the game when your character stops responding to a specific combination.
What the result means
- A key does not light up: it never reached the browser. Before concluding it is broken, try another USB port and clean under it — dirt is the most common cause.
- Rollover 2 or 3: common membrane, no anti-ghosting. For competitive play, that is a real limitation, not an impression.
- Rollover 4 to 6: most keyboards that advertise anti-ghosting. Covers movement combos well.
- Rollover above 6: NKRO or close to it. No in-game combination will hit that limit.
- The test measured less than the box promises: try holding keys from the same area, like WASD. Many keyboards guarantee high rollover only on the gaming keys, not across the whole board.
Where this test stops
- Some keys never reach the browser by operating system decision: the Windows key on its own, Print Screen and the volume and media keys. Not showing up here does not mean they are broken.
- The Fn key on a laptop does not exist for the browser: it is resolved inside the keyboard itself, before becoming an event. What arrives is the key it modified.
- The test measures what the browser receives. A keyboard with its own software and a macro configured may send something different from what you pressed — that is the macro acting, not a defect.
- The drawing on screen follows the Brazilian ABNT2 layout. If your keyboard has another layout, the keys still light up; only the position on the drawing and a few labels may not match.
- The numpad has no drawing of its own, so it does not take up a third of the panel with something laptops lack. Its keys still count toward rollover.
- A wireless keyboard with a weak battery sometimes drops keys intermittently. If the result varies between runs, change the battery before concluding anything.
Common problems
Does it work for a laptop keyboard?
It works the same: the test reads what reaches the browser, and it does not matter whether the key is on an external or built-in keyboard. Two differences are worth knowing. The first is that laptops usually have lower rollover than external keyboards, because the internal space limits the matrix design — if you get 2 or 3, that is normal for the category, not a defect. The second is that the F1 to F12 row on many laptops is set to control brightness and volume: in that case the key does not arrive as F but as a media command, and the drawing does not light up. Hold Fn together to test the actual F function.
A key stopped working. Can it be fixed?
Most of the time yes, without replacing the keyboard. On a mechanical one, the switch for that key is replaced individually and costs little. On a membrane one, the cause is usually dirt or liquid on the trace — worth cleaning before giving up. If several neighboring keys died together, then it is the matrix trace and the repair rarely pays off.
My character stops moving when I press several keys in the game.
That is the classic symptom of low rollover, and this test measures exactly that. Hold the keys you use in the game at the same time and look at the number. If it caps at 2 or 3, the limit is the keyboard and no in-game setting will fix it.
The key lights up here but does not work in the game.
If it reaches the browser, the keyboard is sending it. The problem is between the system and the game: a conflicting shortcut, the manufacturer’s macro software intercepting, or the key not being bound to the action inside the game. Start by checking the key bindings in the game itself.
Do you store what I type?
No. The page only records which physical position was pressed, to light up the drawing, and that never leaves your browser — there is no request to any server. Still, as a basic precaution, do not type passwords on this page.
What next?
Did the test find something odd?
Keyboard in order and the game still responds late? The delay rarely lives in the peripheral: it builds up in the graphics card’s frame queue, in display sync and in Windows interrupt latency. That whole chain is system tuning, not a part swap.