UUID v3 Generator
Derive deterministic v3 UUIDs from namespace + name via MD5 for legacy interop — same inputs, same UUID, entirely on this device.
Determinism
Same namespace + name → same UUID, always
- Namespace UUID
- —
Press ? for shortcuts
Do it offline:$ npx uuid v3 example.com DNS
How to use it
- Choose the namespace that matches your naming scheme, or paste your own namespace UUID.
- Enter the name — the v3 value updates as you type, chip and all.
- Copy the UUID; rerunning the same inputs in any RFC-correct library reproduces it.
UUID v3 derives an identifier by hashing a namespace UUID and a name with MD5. It exists for compatibility: systems that minted v3 ids years ago keep getting identical values today. For new naming schemes RFC 9562 points at v5, and a chip beside every result on this page tells you the same.
How it works
MD5 runs over the namespace UUID's 16 raw bytes followed by the name's UTF-8 bytes; all 16 digest bytes land in the identifier before structure is applied.
Structure costs six bits — version nibble 3, variant 10. The remaining 122 bits are pure digest, so equal inputs can never disagree.
Examples
namespace = DNS · name = example.com9073926b-929f-31c2-abc9-fad77ae3e8ebThe digit after the second hyphen reads 3 — the only visible difference from this pair's v5 sibling. Every result here arrives with the chip “v5 preferred over v3 (SHA-1 vs MD5)”.
namespace = OID · name = 1.3.6.1.4.1ef89b4fd-cc82-39f4-8098-b58dd72a496cDotted object-identifier strings are names too — the OID namespace exists for exactly this shape.
Every example on this page is executed against the same engine that powers the tool, on every build.
Gotchas worth knowing
Names are case-sensitive bytes
Example.com and example.com hash to different v3 values — there is no folding, trimming or normalization before MD5 sees the bytes.
Common errors
Custom namespace is too short — a UUID has 36 characters (8-4-4-4-12), this stops after 8.- A namespace pasted incompletely is refused with the length it actually reached, anchored at the cut.
Name is larger than the 1 MB limit for this tool.- The cap refuses oversized names up front, because hashing a silently shortened name would mint the wrong id.
FAQ
How is v3 different from v5?
Only the digest: MD5 with version digit 3 against SHA-1 with digit 5. Same namespaces, same byte layout, incompatible outputs — the two versions never produce equal values.
When do I actually need v3?
When an existing dataset, protocol or partner system already keys on v3 values — regenerate those ids under any other version and every stored reference breaks.
Does MD5 being broken break v3?
Collision resistance is broken, so a motivated party can craft two names sharing one v3 UUID. Treat v3 ids as labels for trusted inputs, never as tamper evidence.