Windows scenario · NT hash

NT Hash (NTLM) Cracking Calculator

Estimate exhaustive NT hash search on one RTX 5090, then compare other GPUs or cluster sizes.

Offline search of a stolen hash · the table downloads as PNGHow it is calculated →

Attack model

Choose a hash family and hardware profile; the table updates immediately.

Maximum time for a complete brute-force search. Human-chosen passwords can be weaker than the same length suggests.
Length 0-9 a-z a-z, A-Z a-z, A-Z, 0-9 a-z, A-Z, 0-9, ^*%$!&@#
> 10 years (Safe)
1–10 years
1 day – 1 year
1 sec – 1 day
< 1 sec (Cracked)
What matters

Length beats complexity

A long random passphrase usually outperforms a short password with symbols. Use a password manager and unique passwords.

Where this applies

Offline hashes only

The model assumes an attacker already has the hash and can test guesses locally. Online logins behave very differently.

For organizations

Prefer slow hashes

bcrypt, Argon2id, scrypt, and high-cost PBKDF2 make every guess expensive and change the table dramatically.

How to read the estimate

The NT hash is MD4 of the password; Windows stores it in SAM and NTDS.dit. NTLM is more precisely the authentication protocol: its NetNTLMv1 and NetNTLMv2 responses are also in the list and are cracked differently. The NT hash is fast, unsalted and has no work factor, so GPUs test enormous numbers of guesses per second. The preset uses one RTX 5090 and lengths 6–16.

Frequently asked questions

Why is the NT hash so fast to brute-force?

It is a single unsalted MD4 pass, not a slow password-storage function, and it maps efficiently to massively parallel GPU workloads.

How is the NT hash different from NetNTLMv1 and NetNTLMv2?

The NT hash is stored on disk. NetNTLMv1 and NetNTLMv2 are NTLM protocol responses captured on the network. NetNTLMv1 reveals the NT hash through a DES key search for any password; NetNTLMv2 has no such shortcut. Pick the type in the hash list.

Does a long passphrase help?

Yes for random values: length expands the keyspace exponentially, although human phrases may still be vulnerable to dictionaries.

Methodology

Scope of Applicability

Assumptions & Limitations

    References