Password Strength Checker

Score a password by real entropy, see five realistic crack-time models, catch dictionary/sequence/leetspeak patterns, track a 10-point checklist — and generate a strong password or diceware passphrase locally.

Generators Runs in your browser · no upload

Reviewed 17 November 2026 · Free · No account needed

ExecutionYour browserInput never leaves your tab
File retentionNoneNothing is stored
AccountNot requiredNo sign-up, no paywall
PriceFreeNo trial, no usage cap on the tool itself

How to use

  1. Type or paste a password — it is analysed in this tab and never sent anywhere.
  2. Read the verdict, entropy bits, pool size and the five crack-time scenarios from rate-limited login to a GPU cluster.
  3. Work through the requirements checklist and the weak-pattern list (common password, sequence, repeat, leetspeak, year).
  4. Use the generator for a random password (length 8–64, choose character classes, avoid look-alikes) or a diceware passphrase with the word count and separator you want.

Example

Input: P@ssw0rd!

Output: Very weak — common password + letter substitution — cracked instantly on a fast hash

How it works

Strength is estimated from length, the mix of character types, and common patterns (dictionary words, sequences, repeats). A longer password with mixed characters scores higher.

Entropy = length × log2(pool size), where the pool counts only character classes actually used: 26 lowercase, 26 uppercase, 10 digits, 33 symbols. Labels sit at 36, 60, 80 and 100 bits (very weak → very strong). Offline crack time = 2^(bits − 1) ÷ 10 billion guesses per second.

Worked example

Tr0ub4dor&3 uses all four classes (pool 95) over 11 characters → about 72 bits, rated Fair, with an offline crack estimate around 9,000 years.

Edge cases to know

  • A password on the built-in list of 20 most common passwords (password, 123456, qwerty…) is labelled Very weak regardless of its entropy.
  • Warnings flag runs of the same character (aaa), keyboard and number sequences (1234, qwer), digits-only passwords, and anything under 12 characters.
  • The crack estimate assumes one offline attacker at 10 billion guesses per second; online lockouts and distributed attacks are not modelled.
  • No breach database is consulted, so a strong-looking reused password can still have leaked.

Limits & compatibility

  • Entropy assumes uniform random use of the character pool; human-chosen passwords skew weaker than the score.
  • Crack times model five stated attackers (100/hour up to 1 trillion/second) — real hardware varies.
  • No breach database is queried; a strong-looking reused password can still be leaked.
  • The generator uses crypto.getRandomValues but cannot store the result — save it in a password manager.

Privacy

Analysis and generation happen entirely in this tab using local arithmetic and Web Crypto. The password is never transmitted, logged or stored.

Related tools

FAQ

Questions about Password Strength Checker

How is the strength score calculated?

From real entropy. The tool works out the character pool your password actually uses — 26 lowercase, 26 uppercase, 10 digits or 33 other characters — and multiplies the length by log₂ of that pool. The result is labelled very weak, weak, fair, strong or very strong against fixed bit thresholds.

Is my password safe to type into this page?

Yes. The analysis runs entirely in your browser — the password is never transmitted, logged or stored. No breach database is consulted either, on-device or otherwise: nothing is checked against any external list. Bear in mind that a strong-looking password you reuse elsewhere may still have been leaked; this tool cannot see that.

What do the pattern warnings catch?

Structural weaknesses the entropy score alone would miss: a run of the same character such as aaa or 111, keyboard or number sequences like qwerty or 1234, digits-only passwords, and anything shorter than twelve characters. A password on the built-in common-password list is forced to 'very weak' however long it is.

How realistic is the crack-time estimate?

It models one offline attacker trying ten billion guesses per second, using the average case of half the keyspace. That is a fair yardstick for a stolen password file, but online attempts, rate limiting and distributed attacks behave differently. And because human passwords are not uniformly random, real-world odds skew weaker than the number shown.