MD5 Generator

MD5 Generator: Hash Text or a File in Your Browser

This MD5 generator turns any text or any file into its 128–bit MD5 digest — the familiar 32–character hexadecimal string — without sending a single byte anywhere. Type into the box above and the hash updates as you type; drop in a file and it is read in slices and hashed locally, so a multi–gigabyte ISO is handled the same way as a one–line string. There is no upload, no account, no rate limit and no queue.

That last point is worth stating plainly, because many online MD5 tools work by posting your input to a server. This one cannot: the hashing happens in JavaScript inside your own browser tab, so you can paste a configuration snippet or an internal document into it without that content leaving the machine.

How to Generate an MD5 Hash

1. Pick text or file. Text mode hashes exactly the characters in the box, encoded as UTF–8. File mode hashes the raw bytes on disk, which is what a published checksum refers to.
2. Provide the input. Paste or type, or drag a file onto the drop zone. File hashing shows a progress bar; large files stay responsive because the file is read in 4 MB slices rather than loaded into memory whole.
3. Choose an output format. Lowercase hex is the default and is what md5sum, certutil and almost every vendor checksum page uses. Uppercase hex is what Windows’ certutil prints and what some enterprise tools expect. Base64 is the compact 22–character form used in HTTP Content-MD5 headers and in a few package manifests — it encodes the identical 16 bytes.
4. Copy, or verify. Copy the digest with one click, or paste an expected checksum into the verify box to get a straight match / no–match verdict.

Verify a Download Against Its Published MD5 Checksum

Generating a hash is only half the job. The reason vendors publish an MD5 next to a download link is so you can confirm the file arrived intact — that a truncated transfer, a corrupted mirror or a bad disk did not quietly change it. Paste the published value into the verify field above and the tool compares it against the digest it just computed.

The comparison is deliberately forgiving about formatting, because published checksums arrive in several shapes:

• Bare hex — 9e107d9d372bb6826bd81d3542a419d6.
• Uppercase hex — case is ignored on both sides.
• A full md5sum line — 9e107d9d372bb6826bd81d3542a419d6  *installer.iso. Everything after the hash, including the * binary–mode marker, is ignored.
• Base64 — a 22–character string with optional == padding, as seen in Content-MD5 headers, is decoded and compared against the same 16 bytes.

A no–match verdict means the bytes differ, full stop. Before assuming the download is bad, check the two boring causes first: that you copied the entire hash rather than a truncated version of it, and that the checksum you are comparing against belongs to the exact build you downloaded rather than a neighbouring version on the same page.

Command-line equivalents

If you would rather do this in a terminal, or need it inside a script, these produce exactly the same digest this page does:

Note printf '%s' rather than echo in the text row. That distinction causes more confused checksum comparisons than anything else, and it is covered below.

Why Your MD5 Does Not Match: The Three Usual Causes

When a hand–computed MD5 disagrees with one from another tool, the input differed — MD5 itself is completely deterministic. In practice it is nearly always one of these three:

• A trailing newline. echo "hello" | md5sum appends a line feed and returns b1946ac92492d2347c6235b4d2611184. The string hello on its own is 5d41402abc4b2a76b9719d911017c592. Use echo -n or printf '%s' to suppress the newline. The text box on this page hashes exactly what you typed, with no newline added.
• Line endings inside a text file. A file saved with Windows CRLF line endings is byte–for–byte different from the same file saved with Unix LF endings, so the MD5 differs even though the text looks identical. This is the classic cause of a source archive failing verification after passing through a Git checkout with core.autocrlf enabled.
• Text mode versus file mode. Hashing the contents of a file as pasted text is not the same as hashing the file, once encoding, a byte–order mark or a final newline are in play. When you are checking a download, always use file mode.

What an MD5 Hash Actually Is

MD5 (Message–Digest Algorithm 5, published by Ronald Rivest in 1992 and specified in RFC 1321) is a cryptographic hash function that reduces an input of any length to a fixed 128 bits. Those 16 bytes are conventionally written as 32 hexadecimal characters. The output length never changes: an empty string and a 50 GB disk image both produce exactly 32 hex characters.

Because each 512–bit block is folded into a running state, MD5 can be fed incrementally — which is how this page hashes a large file slice by slice without holding it all in memory.

Two properties matter in practice. It is deterministic: the same bytes always give the same digest, on any machine, in any language, forever. And it exhibits the avalanche effect: a one–character change rewrites the entire output. Compare these:

Adding a single full stop changes every one of the 32 characters. That is what makes MD5 useful for spotting corruption: there is no such thing as a “nearly matching” hash, and a single flipped bit anywhere in a download produces a completely different digest.

The empty–string digest d41d8cd98f00b204e9800998ecf8427e is worth recognising on sight. If a tool reports it for a file you expected to have content, the file is zero bytes — the download failed rather than the checksum being wrong.

Is MD5 Secure? Where It Is Broken and Where It Is Fine

MD5 is cryptographically broken, and any page that tells you otherwise is wrong. Wang and Yu demonstrated practical collisions in 2004; by 2008 researchers used a chosen–prefix collision to forge a rogue CA certificate, and today a chosen–prefix collision costs a matter of GPU–hours. CMU’s CERT has described MD5 as “cryptographically broken and unsuitable for further use” since 2008, and NIST disallows it for digital signatures.

Concretely, do not use MD5 for:

• Digital signatures or certificate fingerprints.
• Password storage — MD5 is fast, unsalted by default, and covered by enormous precomputed tables. Use bcrypt, scrypt or Argon2.
• Any integrity check where an attacker chooses or can substitute the input. A collision means two different files with the same MD5, so “the hash matches” stops proving “it is the same file” the moment an adversary is involved.
• HMAC keyed authentication in new designs, and message authentication generally.

Equally, MD5 is entirely legitimate and still widely used for:

• Accidental–corruption detection — verifying a download, a backup or a tape restore against a checksum the vendor published. The threat model here is a flaky mirror or a bad sector, not an attacker.
• Deduplication — finding identical files in a backup set or a media library.
• Cache keys, ETags and shard keys, where the hash is an identifier rather than a security control.
• Malware sample identification — the threat–intelligence ecosystem is still substantially keyed by MD5, so an MD5 remains the practical way to look a sample up in VirusTotal or a vendor feed.

The honest rule: MD5 answers “did these bytes change by accident?” and no longer answers “did someone change these bytes on purpose?” If an attacker gains anything by making two inputs hash the same, reach for SHA–256.

MD5 and SHA-256 Side by Side

If a vendor publishes both, verify the SHA–256. MD5 is worth computing when it is the only checksum on offer, when you are matching against a legacy manifest, or when you are looking a sample up in a database that is keyed by MD5. For the longer comparison including SHA–512, see understanding the differences between MD5, SHA-256 and SHA-512; for the decision framework on legacy systems, see when MD5 is still acceptable.

Common Questions

Can an MD5 hash be decrypted or reversed?

No. MD5 is a one–way function, not encryption — it destroys information, mapping inputs of unlimited length onto 128 bits, so there is nothing to decrypt. What sites advertising “MD5 decrypt” actually do is look the hash up in a large table of previously computed digests of common strings. That works for weak passwords and dictionary words and fails completely for anything else.

Are two different files with the same MD5 possible?

Yes, and that is precisely the problem. By the pigeonhole principle collisions must exist for any fixed–length hash; for MD5 they can now also be constructed on demand, cheaply. Random collisions between unrelated real–world files remain vanishingly unlikely, which is why MD5 still works against accidental corruption.

Should the hash be uppercase or lowercase?

They are the same value. Lowercase is the de–facto standard and what md5sum emits; Windows’ certutil prints uppercase. Comparisons should always be case–insensitive, and the verify box above treats them as equal.

Is there a file size limit?

Not one imposed by this page. The file is streamed through the hash in 4 MB slices, so memory use stays flat regardless of size; the practical ceiling is how long you are willing to wait and what your browser allows. Very large files hash faster from a local disk than from a network share.

Related Tools

• Hash Generator — when you need more than MD5: SHA–1, SHA–256, SHA–384, SHA–512, SHA–3, bcrypt and HMAC, with several algorithms computed side by side over the same input.
• Hash Lookup — when you already hold a hash and want to know what it belongs to, including checks against live malware feeds.
• Hash Type Identifier — when a hash is unlabelled and you need to know whether it is MD5, NTLM, bcrypt or something else, along with the matching hashcat mode.