Typing practice
Type for thirty seconds and find out where your speed went. Every keystroke is timed, so the result names the keys you hunt for, the finger carrying your mistakes and the words you stopped dead on. No account, and nothing leaves this tab.
Pick what to type, type it, then read where the speed went.
- Accuracy
- 0%
- Rhythm
- 0%
- Characters
- 0 / 0
- Time
- 0s
Where the time went
Cost is the time above your own median keystroke, so a long word only appears here if you actually slowed down on it.
- Left pinky
- Left ring
- Left middle
- Left index
- Thumbs
- Right index
- Right middle
- Right ring
- Right pinky
The bar is how much of the run that finger carried. The number beside it is its accuracy, and the time under it is the average gap before its keys land.
Your own keystrokes, at the times you made them. The pauses are the useful part: they are where the next second of speed is hiding.
Your last runs
| Test | Net | Accuracy | Rhythm |
|---|
Kept in this browser only, in local storage. There is no account behind it and nothing is sent anywhere, so clearing your site data clears this too.
How to test your typing speed
Pick what to type
Words, paragraphs, numbers, code, or paste your own text. Then choose a length in seconds or in words.
Click in and type
The clock starts on your first keystroke, not when the page loads. Wrong letters turn red and stay where you left them.
Read where the speed went
Your net speed, then the keys you hunt for, the finger carrying the mistakes, the words you stalled on and a replay of the run.
Using the typing test
Every control on the screen, what it changes, and the five things the tool will not do. The page opens with a thirty-second word test loaded and the words blurred, so clicking in is the only step between you and a result.
What to type: the five sources
The first row of chips picks the text. They are not five flavours of the same list. Each one trains a different part of the keyboard, and the one you want depends on what you are actually slow at.
| Source | What you get, and what it is for |
|---|---|
| Words | The 200 commonest English words, drawn in rough frequency order. The default, and the closest thing to a standard test. Frequency order matters: a list drawn from the top fifty is nearly all three-letter words and flatters your speed. |
| Paragraphs | Full sentences with capitals, commas and full stops. This is the one to use if your real typing is prose, because a word list never asks you to hold shift and never gives you a comma to miss. |
| Numbers | Prices, dates, percentages and phone numbers. The top row and the shifted symbols on it, which is the part of the keyboard almost everybody is slowest on and no word list touches. |
| Code | Real lines of JavaScript, CSS, SQL and shell, brackets and semicolons included. The keys a programmer reaches for hundreds of times a day and never practises. |
| Custom | Your own text, typed or pasted into the box that appears. No length limit, no timer, and it ends at the last word. |
Practising with paragraphs instead of a word list
Paragraphs gives you whole sentences and never cuts one in half. Sixteen of them are written for this tool rather than lifted from a book, because a passage from a novel is somebody else's prose and a passage picked at random out of one is usually missing half the alphabet. Between them these carry every letter and every mark on the main row.
Expect your number to drop by several words a minute against the word list. That drop is the shift key, and it is real: it is there when you write an email too.
The punctuation and numbers toggles
Two buttons, and they only appear beside Words, because nothing else reads them. A punctuation toggle sitting next to a page of code is a control that does nothing.
| Toggle | What it changes |
|---|---|
| punctuation | Decorates about one word in six with a comma, full stop, quote, bracket, apostrophe, semicolon, colon, question mark, exclamation mark or hyphen, and capitalises the word after a full stop. A comma you never meet is a comma you never get faster at. |
| numbers | Drops a bare number into about one word in eight. Lighter than the Numbers source, which is nothing but digits. |
Setting the length
The right-hand chips set how long the run is, in seconds or in words. Picking one clears the other: a test is timed or it is counted, never both.
| Length | When to use it |
|---|---|
| 15 sec | A warm-up, and the fastest number you will ever see. Too short to be a measurement: one stall costs a large share of the run. |
| 30 sec | The default. Long enough to average out a single stumble, short enough to run ten in a row. |
| 60 sec | The classic one-minute typing test, and the length most people mean when they compare numbers. |
| 120 sec | Where stamina starts to show. Most people are slower here than at sixty seconds, and the gap is the useful part. |
| 300 sec | The five-minute test a typing certificate or a pre-employment assessment normally uses. |
| 10 to 100 words | A fixed amount of text rather than a fixed amount of time, so two runs are directly comparable. Useful when you are drilling one thing and want to know whether it got better. |
The typing area
Three lines are visible and the text scrolls a line at a time, so nothing under the box ever moves while you are typing. The words are blurred until you click in, and the timer does not start even then. It starts on your first keystroke.
| What you see | What it means |
|---|---|
| Grey letter | Not typed yet. |
| Dark letter | Typed correctly. |
| Red letter | Typed, and wrong. It stays visible so the mistake is findable rather than swallowed. |
| Red struck-through letter | A letter that is not in the word at all, because you typed past the end of it. |
| Underlined word behind you | A word you submitted with something still wrong in it. |
| The moving bar | The caret. It glides to the next character rather than jumping, and it picks up a halo as the rhythm meter fills. |
Space moves you to the next word whether the current one is finished or not, which is how every serious test behaves and how real typing behaves. Backspace at the start of a word steps back into the word before it, so a typo you notice one word late is still repairable.
The live readout, and the rhythm meter
Four things sit above the text while you type. The first three are the ones every test shows. The fourth is not.
Seconds left
Counts down on a timed run and up on a word-count run, so the big number is always the one you care about.
wpm
Live net speed, using the same definition as the headline on the result. It does not change meaning when the run ends.
% accurate
The share of your keypresses that were right at the moment you made them. Corrections do not scrub the mistake from this figure.
The bar
Rhythm. It fills on evenly spaced correct keys and empties on a stall or a mistake, because evenness is what actually raises a speed.
The four numbers on the result
The Speeds tab shows four different words-per-minute figures for the same run, each with its sum written underneath it. The headline above the tabs is always the net one. This is the part no other typing test will show you, and it is why two tests disagree about identical typing.
The gap is not small. On one measured 26-second run at 95.1% accuracy the same keystrokes scored 93.7, 89.1 and 70.6 words a minute under the three WPM definitions. What each formula is, and what one typo costs at each test length, is measured and dated in the guide to average typing speed.
The chart, and the words that cost you
Two lines. The spiky one is your speed inside each second, and the smooth one is the running net average, whose last point is the headline. A red dot marks a second that contained a mistake, placed on the spiky line so the dip and its cause sit on the same spot.
Underneath, Where the time went ranks the five words that cost you most. Cost is time above your own median keystroke, not raw duration, so a long word only appears if you genuinely slowed down on it. A four-letter word where you stopped dead will outrank a nine-letter word you typed in rhythm.
The Keys tab: your keyboard as a heatmap
Every key you pressed, shaded by how much it is costing you. The accent marks the problem rather than the strength, because a heatmap whose brightest key is the one you already type well is a picture with nothing to do about it.
| Control or reading | What it does |
|---|---|
| Number on a cap | Milliseconds from your previous keystroke to that one. It is the cost of reaching the key, not of pressing it, so a high number means the key took finding. |
| Colour by accuracy | Switches the shading from speed to error rate, and puts a percentage on each cap instead of a time. |
| Pale, faded cap | A key that did not come up in this run. Thirty seconds of common words will not reach z, q or the top row. |
| Hovering a cap | Gives the full row: how many times you pressed it, what share you got right, and the average reach. |
The Fingers tab
The same keystrokes rolled up to the ten fingers of the standard touch-typing assignment. The bar is how much of the run that finger carried, the percentage is its accuracy, and the time is the average gap before its keys land.
The pattern to look for is one finger with a share far above its neighbours and an accuracy below them. That is usually a hand drifting off the home row, and it is the single cheapest thing to fix. Keys nobody is taught a finger for, such as the arrows, are left unassigned rather than guessed at.
The Replay tab
Your own keystrokes, played back at the times you made them. Play, pause, scrub, and slow it to a quarter speed. The pauses are the useful part rather than the typing: they are where the next few words a minute are hiding, and they are almost impossible to notice while you are the one typing.
Your last runs
The panel under the tool keeps the last twenty runs in this browser's local storage, with a sparkline of net speed and the most recent six in full. There is no account behind it and nothing is sent anywhere, which also means it does not follow you to another device and clearing your site data clears it. Clear empties it immediately.
Keyboard shortcuts
| Key | What it does |
|---|---|
| Tab then Enter | Restarts the run with the same text. Tab is not intercepted: it moves focus to the Restart button, which is the next control, so the shortcut costs nothing in accessibility. |
| Space | Submits the current word and moves on, finished or not. |
| Backspace | Deletes a character, and at the start of a word steps back into the previous one. |
| Esc | Leaves full screen. It does not stop a run, because a key that abandons a test you are two minutes into is a key you press once. |
What this tool will not do
Five things, and each of them is a decision rather than a gap.
It will not let you paste into the test
Paste is blocked in the typing field. A test you pasted into measures your clipboard. Pasting into the Custom box, which is a different box, is exactly how that mode is meant to be used.
It will not keep your history across devices
There is no account to hang it on, and adding one would mean a server holding a record of your typing. The runs live in this browser only.
It will not issue a certificate
A certificate is only worth anything if somebody trusts who issued it, and a page that cannot see your run has nothing to attest to. Use the 300-second test to find out what you would score on one.
It will not teach you the home row
There are no graded lessons and no on-screen hands showing which finger to use next. This is a test and a diagnosis, not a course. If you have never touch typed, a structured course will get you further faster, and the Fingers tab here will tell you when it worked.
It will not judge the Fingers tab on a phone
The finger map assumes ten fingers on a physical keyboard. On a touchscreen it will put nearly everything under one or two fingers, which is true and useless. The Keys heatmap and every timing are unaffected.
How this compares to the tests already out there
Seven typing tests were run on 29 August 2026 before this one was built, and the row that matters is the last column: what a signed-out visitor is actually shown after typing.
| Test | Ads on the test screen | Account needed to keep results | Per-key breakdown without an account |
|---|---|---|---|
| floi | None | No account exists | Yes, on the first run |
| monkeytype.com | One slot on the result, plus a cookie banner on arrival | Yes, "sign in to save your result" | No |
| typingtest.com | Three display slots and a video, on the first screen | Yes | No |
| typing.com | Yes | Yes | No |
| keybr.com | Yes, with a paid removal | Yes, for a profile | Yes, but only once it has watched you long enough to calibrate |
| typequicker.com | None seen | Yes | No |
| typespeedtest.com | None | No, kept locally | No |
Two rows deserve reading twice. keybr.com is better than this tool at the thing it does: it generates a lesson aimed at the letters you are worst at and keeps adapting it, which is a genuinely better way to improve than repeating a random word list. If learning is the whole of your problem, use it. monkeytype is better on breadth, with more languages, more themes and a command palette this page has no answer to.
What none of them do is hand a signed-out visitor the diagnosis after thirty seconds. That is the gap this page exists to fill.
Where the keystrokes go
Nowhere. The page is a static file, the timing happens in your own tab, and there is no endpoint to send anything to. That is also why it keeps working with the network off once it has loaded.
The one thing written to disk is the four-number summary of each run, in this browser's local storage, so the history panel survives a refresh. Clear is a button on that panel.
Frequently asked questions
Which words-per-minute number does this page report?
Net WPM: the characters you got right, divided by five, divided by the minutes you actually typed for. It is the same definition the modern browser tests use, which is why the headline here should land near the one you get elsewhere. The other three definitions are on the Speeds tab beside it, each with the sum written out, because a test that shows one number and never says which one it used is the reason two tests disagree about the same typing.
Does a mistake stop the test?
No. Keep going and the wrong letter stays red where you left it, or press backspace and fix it. Backspace at the start of a word steps you back into the word before, so a typo you notice one word late is still repairable. Nothing is ever locked, and there is no minimum accuracy that ends the run early.
Can I practise on my own text?
Yes. Press Custom, type or paste the passage into the box, and press Use this text. There is no length cap and no timer on a custom run: it ends when you reach the last word, because a passage you chose is one you came here to finish. The text stays in the tab, is never uploaded, and is gone when you close the page.
Are my keystrokes recorded anywhere?
Only in the tab you are typing in, and only until you close it. The timings are held in memory to build the result, and the result screen is drawn from them and then thrown away with the page. The one thing that outlives the tab is the four-number summary of each run in Your last runs, which is written to this browser’s local storage and never sent anywhere. Clear is a button on that panel.
Is there a five-minute typing test?
Yes, the 300 chip. That is the length most typing certificates and pre-employment tests use, and it gives a very different answer from fifteen seconds: a sprint flatters you, and five minutes measures the speed you can hold. There is no ten-minute option, because past about five minutes the number stops moving and you are measuring your concentration rather than your typing.
Why does the test not start until I click on it?
Because a clock that starts on page load is a clock that has already been running while you read the page. The words stay blurred until the field has focus, and the timer does not start even then: it starts on your first keystroke. That is also why the number you get is a little higher here than on a test that begins counting the moment it appears.
What does the test do differently on a phone?
Nothing, and that is the problem with one panel of it. The test runs, the on-screen keyboard raises the way it would in any text box, and every timing is real. The Fingers tab is the exception: it assumes ten fingers on a physical keyboard, so on a touchscreen it will pile almost the whole run under one or two of them. That reading is true and useless. The Keys heatmap, the four speeds and the replay are all unaffected.
What is the rhythm meter measuring?
How evenly spaced your keystrokes are, not how fast. It fills while correct keys land at a steady interval and empties on a mistake or a pause. It is there because evenness is the thing that actually raises a typing speed, and because a bar that responds to your hands is more use mid-run than a number you cannot read while typing. It is not saved and it feeds no score.