NanoID Generator

Shorter than a UUID, comparable strength

A default Nano ID is 21 characters from a 64-symbol alphabet, giving about 126 bits of entropy — marginally more than a UUIDv4's 122, in 21 characters instead of 36. The saving comes from using a larger alphabet rather than from weaker randomness.

URL-safe by construction

The alphabet is A–Z, a–z, 0–9, hyphen and underscore, so an ID needs no encoding in a URL, a filename or an HTML attribute. UUIDs are also URL-safe but carry hyphens in fixed positions and are visually harder to distinguish at a glance.

The alphabet is configurable, and that has consequences

Shortening the ID or removing ambiguous characters reduces entropy, and the reduction is exponential. A 10-character ID from a 36-symbol alphabet has about 52 bits, which collides at roughly 250 million IDs by the birthday bound — fine for short-lived links, wrong for a primary key. Any custom configuration needs its collision probability checked rather than assumed.

Random IDs fragment database indexes

Like UUIDv4, Nano IDs scatter inserts across a B-tree rather than appending, causing page splits and index fragmentation at scale. Where the ID is a clustered primary key, a time-sortable identifier such as ULID or UUIDv7 avoids that while keeping unguessability.

Unguessable is not authenticated

A Nano ID in a URL is opaque enough to resist enumeration, which makes it suitable for share links. It never expires and appears in logs, referrer headers and browser history, so anything requiring revocation or a lifetime needs a real token.