Password Strength Tester (Entropy Bits)

Estimate password entropy in bits and the time an offline brute-force attacker would need at common cracking rates. Pure client-side — your password is never sent anywhere.

Inputs

Type or paste the password. Computation is local-only — nothing is uploaded.

NIST SP 800-63B treats 10^10/s as the modern offline baseline. Bcrypt/scrypt/Argon2 hashes drop that by 4-6 orders of magnitude.

Result

Loading calculator…

General calculation reads

Amazon affiliate

As an Amazon Associate we may earn from qualifying purchases. This does not add cost for you.

How to use this calculator

  • Type or paste the password into the input.
  • Choose the attacker rate — 10 G/s is the realistic modern baseline; pick higher for nation-state threat models.
  • Read the entropy bits and the mean / worst-case crack-time. Flags below the result indicate why the number may be over-stated.
  • Aim for ≥ 80 bits for accounts that matter; ≥ 128 for master passwords or encryption keys.

About this calculator

A password’s strength is best expressed as its entropy in bits — log₂ of the number of equally-likely passwords an attacker has to try. For a random 12-character string mixing upper, lower, digit, and symbol, that is about 78 bits — strong enough to resist a single GPU rig for centuries on a non-cached hash. This tester computes the theoretical entropy from password length and the character classes it contains, then converts to mean crack-time at four common attacker speeds: rate-limited online, slow bcrypt offline, modern GPU, and state-actor cluster. It also flags weaknesses that ruin the entropy estimate: dictionary words, short length, and use of only one character class.

How it works — the formula

entropy_bits = length · log₂(pool_size) mean_crack_time = (2^entropy_bits / 2) / attacker_rate

For a uniformly random password from a pool of N characters, the number of equally-likely possibilities is N^length, so log₂ gives the entropy in bits. Mean-case search costs half that. Real-world passwords are far less random; entropy bits are an upper bound on resistance to brute force, not on resistance to dictionary attacks.

Worked examples

Example 1
"password" (lower-only, 8)
Inputs:
pwd=password
Output:
~37.6 bits but dictionary-flagged → minutes
Example 2
12 random mixed-case + digit + sym
Inputs:
pwd=Tr0u4!H#9Vk2
Output:
~78 bits → ~years on a GPU rig
Example 3
20-char Diceware passphrase
Inputs:
pwd=correct horse battery staple
Output:
~92 bits → centuries on a GPU rig (but only ~52 bits if attacker uses a word-list)

Limitations

  • Entropy ≠ real security: dictionary + rule-based crackers find "high-entropy looking" passwords in seconds when they follow common patterns.
  • No breach-list check — use haveibeenpwned.com/Passwords for that.
  • Attacker-rate selection is a coarse model — real cracking speed depends on the specific hash function and hardware.

Educational estimate only — for compliance use a tool that also checks against breach databases and known dictionaries.

Frequently asked

Why is the actual security usually less than the entropy bits?+
Entropy bits assume every character was chosen uniformly at random from the available pool. Humans almost never do that — they pick common words, predictable substitutions ("P@ssw0rd"), and keyboard runs. Real-world crackers use dictionary + rule sets that exploit those patterns, recovering many "high-entropy looking" passwords in seconds.
What is NIST’s recommended length?+
NIST SP 800-63B (2024 revision) recommends a minimum of 8 characters for memorised secrets and 15+ for high-assurance accounts, with no required composition rules. Length is the dominant factor — a 16-character lowercase passphrase beats an 8-character mixed-class password.
Why does the attacker rate matter so much?+
Crack-time is inversely proportional to attacker speed. A 64-bit password resists 1k/s for ~290 million years but falls to 10G/s in ~3 weeks. Always pick a rate appropriate to your threat model — if you assume the service hashes with bcrypt (cost 10+), drop to 1M/s; if MD5/SHA1/NTLM (legacy), use 10G/s.
Why are passphrases recommended over complex short passwords?+
Four random dictionary words from a 7,776-word list (Diceware) is ~52 bits — comparable to an 8-char fully-random ASCII password but vastly easier to memorise. Length is multiplicative in entropy; complexity is additive. The xkcd cartoon "correct horse battery staple" makes this point.
Is my password sent anywhere?+
No. All computation runs locally in your browser. The password is never uploaded, logged, or stored. You can also disconnect from the network and the tool still works.
What about pwned-password breach lists?+
This tool checks character composition only, not breach databases. For a "has this password been pwned?" check use haveibeenpwned.com/Passwords (k-anonymous API — your password is hashed locally and only a prefix is sent).

Related calculators

More tools you might like

Hand-picked tools that pair well with this one — same audience, same intent.