Password Strength Checker

See how many guesses a password really costs, and how long it would survive each kind of attack.

Runs on your deviceSecurity & PrivacyWorkingNo account needed

What this tool does

Password Strength Checker does not count characters and award points for variety. It reads the password from left to right and takes the longest recognisable pattern at each position: entries from a built-in list of 195 very common passwords and words from a 1,289-word dictionary, either spelled out or with l33t substitutions such as 4 for a or $ for s; ascending and descending runs; walks along a keyboard row; dates and four-digit years; and blocks that simply repeat. Each matched segment is priced at the guesses it costs someone whose software already knows that pattern, every unmatched character is priced at the size of its own character class, and the segment prices are multiplied together with a further factor for the order the segments could have been assembled in. The total is then held at or below plain brute force over the whole password, so the segmentation only ever lowers the estimate.

The 0 to 4 rating is deliberately the lower of two views. One is that raw guess count. The other is how long the password would last if the service storing it leaked its database and the hashes were attacked on GPUs at ten billion guesses a second — anything falling in under an hour is capped at Weak, under a year at Fair, under a century at Strong. Without that cap a password that dies in an afternoon of GPU work would carry a green Very strong label, which is precisely the reassurance people act on. The report lists all four scenarios one under the other: rate-limited online guessing at 100 an hour, unthrottled online guessing at 10 a second, bcrypt at 10,000 a second, and the fast-hash GPU figure. The same password is comfortable on the first line and finished on the last.

The honest limit is the wordlist. Fewer than 1,500 built-in entries stand in for what real cracking software carries: millions of leaked passwords, plus names, places, lyrics and film titles across dozens of languages, all run through mangling rules. A password built from an uncommon English word, a word in another language, or anything drawn from your own life — a surname, a pet, a postcode — can therefore be reported as strong when a targeted attack would reach it quickly. Where a password under 16 characters rates Fair or better with nothing at all recognised inside it, the report prints that caveat beneath the score; a longer one is left to stand on its length. There is also no connection to any breach corpus, so it cannot tell you whether this exact string has already been published in a leak. If you have used it anywhere before, treat the rating as void.

Nothing you type is transmitted. The matching, the entropy arithmetic and the crack-time table are plain JavaScript running in this tab, no request is made while it works, and nothing is kept once you leave — which is why the page still functions with the network switched off, and why a password you are actively using is safe to paste in. The report masks it by default, printing bullets rather than characters, so a screenshot or a glance over your shoulder gives nothing away; the Show the password in the report checkbox reveals it, along with the unmatched random stretches in the pattern table.

How to use it

  1. Type or paste the password. One password per check. A trailing newline is trimmed, and anything longer than 256 characters is refused rather than analysed, since length alone has already put it out of reach.
  2. Decide whether to show it. Show the password in the report is off by default, so the report prints bullets. Turn it on when you need to see exactly which characters formed each matched pattern.
  3. Read the pattern table first. It lists every segment found, the kind of pattern it is, and the guesses that segment costs. The cheap rows are the ones holding the score down, and they are what to change.
  4. Compare the four attack speeds. The crack-time table puts the same password against rate-limited login attempts and against GPUs on a leaked hash, timing the expected half of the search space rather than all of it. Judge it by the bottom line, not the friendliest one.

Processing & privacy

Runs on your device. Your file is processed directly in this browser.

Common use cases

Testing a password you already use

Since nothing is transmitted, a live password can be pasted here without that act becoming the risk. The pattern table usually explains why a familiar one is weaker than it feels.

Showing a colleague why a rule is not enough

Summer2024! satisfies most corporate complexity policies and is still an entry from the common-password list followed by a recent year. Both parts are named and priced in the table.

Sanity-checking a passphrase

Several unrelated words score well; a single word dressed up with substitutions and a symbol does not. The segmentation shows which of the two you actually built.

Choosing a minimum length for a policy

Try candidate lengths and watch where the offline fast-hash line stops being measured in hours. That argues for 16 characters more convincingly than a line in a staff handbook.

Questions

Is my password sent anywhere?
No. Wordlist matching, entropy and the crack-time arithmetic all run in this browser tab, no network request is made, and nothing is stored between visits. Disconnect from the internet and the tool behaves identically.
Does it check whether my password has been breached?
No. There is no lookup against any leaked-password corpus, so a string that appears in every cracking dictionary in circulation can still be rated highly here if its shape happens to be unusual. If you know a service holding this password was breached, Breach Exposure Guide covers what to do next.
Why does it rate my password lower than other checkers do?
Because the label is capped by the offline attack. A password costing around 10^14 guesses is respectable against a login form and only a few hours of GPU work against a leaked MD5 hash, so a guess count that would otherwise read Very strong is reported as Fair, with a line naming the downgrade and the time behind it.
What does the entropy figure mean?
It is the base-2 logarithm of the estimated guess count — password entropy in bits, where each additional bit doubles the work. Where the patterns found have shrunk the search space by more than five bits, the report also prints what a naive character count would have claimed instead.
Is a Very strong rating a promise?
No, it is the strongest thing this analyser can say with the lists it holds: whatever patterns it did find still leave a guess count that survives ten billion guesses a second for more than a century. It cannot account for a real wordlist containing your exact password, nor for phishing, keylogging or reuse. A unique password per site and two-factor authentication do more than any rating.
Can I check a whole list at once?
No. The input is treated as one password, so a pasted list would be analysed as a single long string, and past 256 characters it is refused. Check them one at a time, or use Password Generator to produce replacements.

Related tools

  • Protect PDF — Add a password and set permissions on a PDF.
  • Sign PDF — Draw, type or upload a signature and place it on the page.
  • Hash Generator — Compute MD5, SHA-1, SHA-256 and SHA-512 digests.
  • Password Generator — Generate strong passwords with rules you control.
  • Redact PDF — Permanently remove sensitive content, not just cover it.
  • Remove PDF Metadata — Strip identifying metadata before you share a file.