Online mouse test

Buttons / wheel / hold / double-click Measured directly
Which button, held how long, how far apart two clicks landed, how far the wheel turned — all on one card.

Press a mouse button, or roll the wheel
Presses—
Held now—
Click gap—ms
Hold time—s
Notches—
Last 10
When a click gap is under 90ms the counter for that button turns red — no hand moves that fast. Whether a hold broke mid-way is invisible here: that needs a whole drag to judge the shape, so it lives on the hold & drag page. The two wheel slots: direction is the wheel and the arrow, and the notch count is an estimate — the browser never says how big a "notch" is. The strip is one cell per scroll event: down sinks, up rises, a red cell is a back-glitch. A zero in the side-button slot only means the page never heard it — on some mice the side buttons do not arrive as mouse events at all.Separate pages:Drag · Double · Wheel
Mouse CPS test Measured directly
Measure clicks per second in 5, 10 and 30 second runs, with a per-second chart and your personal best.
Click fast in here
Instant CPS—
Clicks—
Five seconds in a card is enough for a rough idea. For 10 / 30 seconds, a per-second chart and a personal best, open the page.
Mouse trail test Shape only
Draw a stroke and watch the raw movement: smoothness, jitter, dropouts, and whether this mouse and mousepad get along.
Draw a stroke in here
Samples—
Isolated spikes—
A sensor dropping points and a fast flick look identical in the data, so this reports shape only — never "how many frames were lost".
Mouse polling rate test Measured directly
Measure the polling rate from 125 Hz to 8000 Hz — the real figure, the peak, and how much it wobbles second to second.
Move back and forth quickly in here
Polling rate—Hz
Samples—
How fast you can sweep depends on inches-per-second × DPI, and that does not fit in a small box — so the number here is only a lower bound. To measure it properly, use the full-page version.
Mouse DPI test Estimated
Estimate the real DPI: push the mouse a fixed physical distance and the screen movement is divided by it. Pointer acceleration has to be off.
This one needs a ruler pushed along a fixed physical distance, then a back-calculation from screen movement. There is no room for that ruler in a card.
Your hand, the pointer-speed slider and display scaling all end up in the result, so the fixed distance has to be pushed exactly as instructed.
Mouse acceleration test Estimated
Move the same physical distance at two different speeds and compare. If fast movement travels further, pointer acceleration is on.
This one runs two passes at different speeds and compares them; a card is not big enough for that.
The comparison is "same physical distance, different speeds", so both passes have to be finished before there is a conclusion.
Keyboard key test Measured directly
Every key lights up the moment it is pressed — useful for spotting a dead key or a set of keys that cannot be held together.
Keys pressed on the home page scroll the page and collide with browser shortcuts, so press-to-light-up only works on that page.
It uses the physical position of the key, so input method, letter case and modifier keys cannot affect the reading.

More tools

The cards above are diagnostics — they answer "is this mouse broken?". The ones below are for fun. Scores stay on your machine either way, and each of them says plainly which numbers are solid and which are not.

Utilities

Nothing in this section measures the mouse: audio and camera, plus hash, JSON and unit conversion that simply compute. None of it measures anything, so there is no accuracy to report — each page says whether it views, converts or calculates.

Every item in the toolbox carries a one-line description and a confidence label, so you can tell what it measures and how far to trust it.

About these tests

Everything runs from the page itself — no driver, no download, no client. Buttons, polling rate, sensor and wheel are all measured in your browser, and nothing leaves your machine.

Each card carries a badge in the top right saying how far its numbers can be trusted. To find out whether a button is failing, use the ones marked "Measured directly". For the ones marked "Shape only", read the line at the bottom of the card — the clues they give you are real, but a clue is not proof. The four tiers:

Measured directly: The browser hands over the raw event stream; what is counted is what happened.
Estimated: A conversion or a back-calculation sits in between. If the assumption fails, the result fails with it.
Shape only: Only the shape is trustworthy — it cannot support a conclusion like “frames were dropped”.
Direction only: The direction bit is the only thing that needs no assumption; everything else is estimated.