Short answer: pick v4 for unpredictable identifiers, v7 when the UUID is a database key that should sort by creation time. Set the count, choose case and hyphens, and the list is generated in this tab — copy it or download it as a .txt file.
v4 versus v7 in one paragraph each
v4 spends 122 of its 128 bits on randomness. It is the safest choice when an identifier must not be guessable, such as a public share link. The cost is that inserts land all over a B-tree index, which fragments it once the table is large.
v7 starts with the Unix time in milliseconds (48 bits), then a version nibble, 12 more random bits, the variant and 62 more random bits. Because the prefix grows with time, new rows append to the end of an index instead of splitting it. The trade-off is that the creation time is readable from the identifier — which is fine for a database key and not fine for a secret token.
What this page does not do
- It does not guarantee monotonic values inside a millisecond.
- It does not generate v1, v3, v5 or v8 — those need a MAC address, a namespace or a hash.
- It does not check the values against your database; uniqueness is probabilistic by design.
- It does not transmit anything, so there is no server-side audit trail of what you generated.
Working with identifiers in a payload? The JSON formatter will pretty-print the document, and the text diff shows what changed between two fixtures.