Password Generator
Generate strong, random passwords with custom length and character rules. 100% client-side using crypto.getRandomValues(). Never sent to any server. Free.
Password Generator Tool
NIST SP 800-63B-4 §3.1.1.2 requires verifiers to enforce a minimum of 15 characters for a password used on its own, and allows a minimum of eight only when the password is one factor of multi-factor authentication. The 8-character floor on this slider is for systems that force your hand — anything under 15 is a compatibility setting, not a recommendation.
Each result shows its entropy in bits — length × log₂(pool size), where the pool is the character set you enabled (26 + 26 + 10 + 26 = 88 with everything ticked). Bands: under 50 bits Weak, 50–70 Fair, 70–90 Good, 90–120 Strong, 120 bits and above Very Strong. The "time to crack" under each password is the average for an offline attacker trying 100 billion guesses per second against a fast, unsalted hash — a deliberately pessimistic benchmark. A properly salted, slow hash (bcrypt, Argon2) would take far longer.
crypto.getRandomValues(). Never sent to any server.
How to Use the Password Generator
Set your password length
Drag the slider to choose between 8 and 64 characters. The default is 20. NIST SP 800-63B-4 §3.1.1.2 requires verifiers to enforce a 15-character minimum for a password used as a single factor, so treat 15 as the floor and only drop below it when a site refuses longer input. For banking, email and your password-manager master password, 24 or more costs you nothing when the manager does the typing.
Choose character types
More character types means a stronger password. Enable symbols for maximum security. If a site does not allow symbols, uncheck that option — the generator will adapt. Enable "Avoid ambiguous characters" if the password will be typed by hand rather than pasted.
Click Generate
Each click creates fresh random passwords using your browser's cryptographic random number generator. The strength meter reports the actual entropy of each result — length × log₂(pool size) — in bits, then bands it: under 50 bits Weak, 50–70 Fair, 70–90 Good, 90–120 Strong, 120 and above Very Strong. Re-generate as many times as you like — it is instant and free.
Copy and use your password
Click Copy next to any password to copy it, then paste it into your password manager or account registration form. Use Copy All to grab multiple passwords at once, one per line. Store every password in a password manager — never reuse across accounts.
Password Security in 2026 — Why Length and Randomness Are Everything
What Makes a Password Actually Strong
The single most important factor in password strength is length — not complexity. A 12-character password chosen randomly from a large character set is significantly harder to crack than an 8-character password that mixes uppercase, numbers, and symbols. The reason is entropy: the amount of information (measured in bits) packed into the password. Entropy is calculated as bits = log₂(charset_sizelength). Doubling the character set adds one bit per character; adding one character always adds the same logarithmic gain — making length the dominant lever.
This is why "P@ssw0rd" — which looks complex — is cracked in seconds by any modern tool. Attackers use dictionary-based rules that substitute letters for numbers and symbols (called "mangling rules"). NIST's Digital Identity Guidelines take the same view: SP 800-63B-4 §3.1.1.2 states that verifiers and credential service providers "SHALL NOT impose other composition rules (e.g., requiring mixtures of different character types) for passwords", and the same list of requirements tells them not to force periodic password changes unless there is evidence the credential has been compromised. Those two rules existed for decades and produced short, predictable passwords — Password1! in January, Password2! in April.
The same section requires verifiers to compare every new password against a blocklist of "commonly used, expected, or compromised" values — breach corpuses, dictionary words, the name of the service. That check is the reason human-chosen passwords keep failing: they come from a small, guessable distribution. A password drawn character-by-character from a cryptographic random source, as this tool does, has no pattern to mangle and no reason to appear on any such list.
How This Tool Measures Strength — and Why the Number Is the Honest Part
The meter next to each password reports entropy in bits, computed as length × log₂(pool size). The pool is the character set you actually enabled: 26 uppercase + 26 lowercase + 10 digits + 26 symbols = 88 characters with every box ticked, so each character carries log₂(88) ≈ 6.46 bits. A 16-character password from the full set is therefore 16 × 6.46 ≈ 103 bits; dropping symbols shrinks the pool to 62 and the same length to 16 × log₂(62) ≈ 95 bits. Ticking "avoid ambiguous characters" removes five characters from the pool (88 → 83), costing about 0.08 bits per character — a rounding error next to one extra character.
Length dominates, and the number makes that obvious: a 64-character password using nothing but lowercase letters is 64 × log₂(26) ≈ 301 bits — nearly three times the 103 bits of the "all four character types" 16-character password that looks stronger. Any meter that awards points for having an uppercase letter will rank those two the wrong way round. The bands here are drawn on the entropy figure alone: under 50 bits Weak, 50–70 Fair, 70–90 Good, 90–120 Strong, 120 bits and above Very Strong. Entropy is a property of how the password was generated, not of how it looks — which is also why the figure only tells the truth for passwords this tool produced at random.
"Verifiers and CSPs SHALL NOT impose other composition rules (e.g., requiring mixtures of different character types) for passwords." — NIST SP 800-63B-4 §3.1.1.2
Why Password Managers Are Essential in 2026
Password reuse is what turns one company's bad day into yours. When a site is breached, attackers immediately replay the stolen credentials against banking, email, social media and cloud storage. This technique, called credential stuffing, is fully automated and runs at massive scale — and it only works because the same password opens more than one door. Unique random passwords break the chain: a breach at one site leaks exactly one account.
A password manager is the only practical way to keep a different high-entropy password for every account. NIST is unusually direct about this: SP 800-63B-4 §3.1.1.2 states that "Verifiers SHALL allow the use of password managers and autofill functionality", and that verifiers should permit pasting into password fields, noting that password managers have been shown to increase the likelihood that people choose stronger passwords — particularly when the manager includes a generator. If a site blocks paste, that is the site failing the guideline, not you failing at security.
Solid options in 2026: Bitwarden (open-source, free tier), 1Password (strong enterprise features), and KeePass (fully offline, open-source). Generate the password here or in the manager itself, store it there, and never type it again. The one password you do have to memorise — the manager's master password — is the case where a long passphrase beats a random string, because you have to reproduce it from memory on a phone keyboard at an airport.
10 Facts About Password Security
This generator's full character pool is 88 characters (26 + 26 + 10 + 26), so every character you add is worth log₂(88) ≈ 6.46 bits of entropy.
Length beats variety, and the arithmetic proves it: 16 characters from all four sets = 103 bits, while 64 lowercase-only characters = 301 bits.
81% of hacking-related breaches involved stolen or weak passwords — a finding from Verizon's 2017 Data Breach Investigations Report, still the most-quoted figure in the field.
NIST SP 800-63B-4 §3.1.1.2 tells verifiers they "SHALL NOT impose other composition rules" — the forced uppercase-plus-digit-plus-symbol rule is now the opposite of best practice.
The same NIST section forbids routine password expiry: verifiers "SHALL NOT require subscribers to change passwords periodically", and force a change only on evidence of compromise.
Password reuse is the #1 security vulnerability — if one site is breached, attackers test those credentials on banking, email, and other services automatically.
crypto.getRandomValues() uses your device's OS-level cryptographic random number generator — far more secure than Math.random() which is predictable.
A passphrase like "correct horse battery staple" (4 random words) has more entropy than "P@ssw0rd1!" despite looking simpler.
Bitwarden, the open-source password manager, was audited by cybersecurity firm Cure53 in 2023 — making it one of the most transparent options available.
A six-word Diceware passphrase drawn from the standard 7,776-word list (6⁵ — five dice per word) carries 6 × log₂(7776) ≈ 77.5 bits.
Frequently Asked Questions
-
Very secure. This generator uses
window.crypto.getRandomValues()— the same cryptographic-grade randomness used by browsers for TLS and other security-critical operations. All generation happens entirely in your browser; no data is ever sent to any server. The generated passwords never appear in any log or database. -
Math.random()is a pseudo-random number generator (PRNG) — it produces numbers from a deterministic algorithm seeded by a small internal state. With enough outputs, an attacker can reconstruct the seed and predict future values.crypto.getRandomValues()draws entropy from your operating system's hardware-backed random source (e.g. thermal noise, interrupt timing), making prediction computationally infeasible. For passwords, only cryptographic randomness is acceptable. -
NIST SP 800-63B-4 §3.1.1.2 sets the floor: verifiers "SHALL require passwords that are used as a single-factor authentication mechanism to be a minimum of 15 characters in length", and may allow eight only when the password is one factor of multi-factor authentication. This tool defaults to 20. For high-value accounts (banking, email, password manager master password), 24–32 characters costs nothing when a manager does the typing. The same section says verifiers "SHOULD permit a maximum password length of at least 64 characters" — which is why this slider stops at 64.
-
Yes, when the site allows it. Including this tool's 26 symbols expands the pool from 62 characters (upper + lower + digits) to 88, which is log₂(88) − log₂(62) ≈ 0.51 extra bits of entropy per character. Over 16 characters that is 103 bits with symbols versus 95 without — real, but smaller than the 6.5 bits you gain by simply adding one more character. Some sites restrict which symbols are accepted — if a generated password is rejected, uncheck symbols, add a few characters of length, and regenerate.
-
Entropy, in bits: length × log₂(pool size), where the pool is the character set you enabled (88 characters with all four boxes ticked, 83 if you also avoid ambiguous characters). The number shown next to each password is that figure — 16 characters from the full set is ≈103 bits, 20 characters is ≈129 bits, 8 characters is ≈52 bits. The label bands it: under 50 bits Weak, 50–70 Fair, 70–90 Good, 90–120 Strong, 120 and above Very Strong. Because entropy depends on how a password was generated rather than how it looks, this figure is only meaningful for the random passwords this tool produces — no meter can measure the entropy of a password a human chose.
-
No. Generated passwords exist only in your browser's memory for as long as you have the page open. They are never sent to RECATOOLS servers, never logged, and never stored in cookies, localStorage, or any other persistent storage. Closing or refreshing the page discards them permanently. This is by design — store your passwords in a dedicated password manager such as Bitwarden or 1Password.
-
Some characters look similar in many fonts: zero (0) and uppercase O, lowercase L (l), the number one (1), and uppercase I all look nearly identical in many typefaces. If you ever need to read or type a password by hand — for example, entering it on a TV remote or a device where paste does not work — enabling this option removes those characters from the pool, eliminating transcription errors. If you always paste passwords, leave this unchecked for maximum entropy.
-
A password manager is an application that generates, stores, and auto-fills unique random passwords for every account you use. You remember only one master password; the manager handles everything else. This eliminates password reuse — the #1 cause of account takeovers. Yes, you should use one. Recommended options: Bitwarden (free, open-source, cross-platform), 1Password (excellent UX, subscription), KeePass (fully local, no cloud). NIST SP 800-63B-4 §3.1.1.2 states that "Verifiers SHALL allow the use of password managers and autofill functionality" — a site that blocks paste into its password field is breaking the guideline.
-
A passphrase is a sequence of randomly chosen words — for example, "correct horse battery staple" (popularised by xkcd #936). Four random common words produce approximately 44 bits of entropy using a 2,000-word dictionary, and six words produce around 66 bits. Passphrases are easier to memorise than random character strings, making them ideal for master passwords that you must remember. For accounts stored in a password manager, a random character password of 16+ characters from this tool provides more entropy and is the better choice.
-
Yes — a randomly generated 16-character password from this tool's 88-character pool has ≈103 bits of entropy (16 × log₂(88)), which is far beyond what an offline guessing attack can search, and that is before the site's own password hashing scheme makes each guess expensive. We deliberately do not print a "years to crack" figure: it depends entirely on the attacker's hardware and on which hashing scheme the site used, so any single number is marketing rather than measurement. If you want headroom, use the 20-character default or push to 24 — 20 characters is ≈129 bits. It costs nothing with a password manager, and this generator makes it instant.
Related News
You may be interested in these recent stories from our newsroom.
-
One Generated Prompt Beat Up to 63 Per Cent of a Top Conference's LLM Techniques
A University of Virginia team reproduced 35 ICSE 2026 techniques and outperformed 37 to 63 per cent of them with a single prompt to a newer...
-
The Sixth Exploited Langflow Flaw This Year Is Not a Vulnerability Story
VulnCheck has recorded 360 attacks on a Langflow flaw fixed in January. It is the sixth separately identified Langflow vulnerability exploit...
-
Broadcom Already Maintains Spring and RabbitMQ. Now It Will Sell You Clean Builds.
A reproducible build proves the binary matches the source. It says nothing about whether the source is safe, which is where the damaging inc...
Pick up where you left off
Stored only in this browser — never sent to our servers.