What is the difference between v4 and v7?

v4 is fully random — 122 bits of entropy with no internal structure. v7 (RFC 9562) embeds a millisecond Unix timestamp in the leading 48 bits, so values sort lexicographically by creation time. Use v4 for opaque identifiers; use v7 when you want time-ordered keys for database indexes.

Are these cryptographically secure?

Yes. Both v4 and v7 generation use crypto.getRandomValues() from the WebCrypto API, which draws from your operating system's CSPRNG. Math.random() is never used.

How likely is a collision?

For v4: practically zero. You would need to generate ~2.71 quintillion UUIDs to have a 50% chance of one collision. For v7 within the same millisecond, randomness covers the remaining 74 bits, so collisions in practice are also negligible at any normal volume.

Can I generate them in bulk?

Yes. Set the count and the tool will produce up to a million values at once, with options to copy them as a list, JSON array, CSV, or SQL VALUES clause.

Is a UUID truly unique? What are the chances of a collision?

Yes, for all practical purposes. A standard UUIDv4 uses 122 random bits, providing 5.3 x 10^36 possible combinations. To put this in perspective, you would need to generate 1 billion UUIDs per second for about 85 years to reach a 50% probability of a single collision.

What is UUIDv7 and why is it better for databases?

UUIDv7 combines a Unix timestamp (precise to the millisecond) with random data. Because the timestamp comes first, UUIDv7 strings are sequentially sortable in chronological order. This makes them significantly better for database primary keys than UUIDv4, as sequential writes prevent B-Tree fragmentation and eliminate indexing hot-spots.

Free Online UUID & GUID Generator

Generate UUIDs in v1, v4, v5, and v7 formats. Validate, bulk generate up to 5000, natively format, and analyze timestamps.

Count (1–5000)

What is the difference between a UUID and a GUID?

There is absolutely no functional or structural difference. A UUID (Universally Unique Identifier) is the standard terminology defined by RFC 4122 and is used widely across web, Linux, and Apple ecosystems. A GUID (Globally Unique Identifier) is simply Microsoft's marketing term for a UUID, used primarily in the Windows, .NET, and COM ecosystems. Both generate 128-bit values formatted identically.

Is a UUID truly unique? (Collision Probability)

Yes, for all practical purposes. A standard UUIDv4 uses 122 random bits (with 6 bits reserved for version/variant data), which translates to exactly $2^122$ (or 5.3 x 10³⁶) possible combinations.

To comprehend how vastly improbable a collision is: you would need to generate 1 billion UUIDs every second for approximately 85 years just to reach a 50% probability of a single duplicate ever occurring. Relying on UUIDv4 uniqueness for database primary keys or transaction IDs is mathematically sound and universally accepted in software engineering.

Why is UUIDv7 better for modern databases?

UUIDv4 is entirely random. While great for obfuscation, using random strings as primary keys in traditional relational databases (like PostgreSQL or MySQL) causes massive performance degradation known as B-Tree Index Fragmentation. Because the strings are out of sequential order, databases must constantly rewrite index pages globally whenever new entries are inserted.

UUIDv7 solves this by dedicating the first 48 bits to a high-precision Unix chronological timestamp, followed by random entropy. The result is a UUID that remains universally unique but natively sorts chronologically. Databases can seamlessly append new rows sequentially to the B-Tree index, drastically reducing write amplification and heavily optimizing massive table insertions.

About UUID Generator

A UUID is a 128-bit identifier that is unique without a central registry — useful whenever you need to mint a primary key, request ID, or external reference without coordinating with any other system. Different versions trade off predictability against sortability. v4 is fully random and the safe default. v7 is timestamp-prefixed and sorts in insertion order, which is much friendlier to B-tree database indexes.

What this tool does

v4 generation — fully random, RFC 9562 compliant.
v7 generation — Unix-millis prefix plus 74 random bits; sorts by creation time.
Bulk output — up to a million UUIDs at once.
Format options — hyphenated, no-hyphen, uppercase, braces, URN.
Export — plain list, JSON array, CSV, or SQL VALUES rows.

When to pick which version

Pick v7 if the UUID will be a database primary key — time-ordered values keep index pages packed and avoid the random-insert fragmentation that plagues v4 on large tables. Pick v4 when you want IDs to be opaque and to leak nothing about creation time (anti-enumeration in public URLs, session tokens, idempotency keys). Avoid v1 — it leaks the MAC address. Avoid v3 and v5 unless you genuinely need deterministic, namespace-based IDs. v6 exists but most ecosystems standardized on v7.

Pipeline

UUIDs commonly pair with:

Hash Generator — produce a deterministic SHA digest of a UUID for sharding or namespacing.
Password Generator — generate stronger random secrets when entropy matters more than format.

Privacy

Generation runs entirely in your browser using WebCrypto. No server is contacted. Generated values are never logged or transmitted. Read our privacy policy.

Frequently asked

What is the difference between v4 and v7?: v4 is fully random — 122 bits of entropy with no internal structure. v7 (RFC 9562) embeds a millisecond Unix timestamp in the leading 48 bits, so values sort lexicographically by creation time. Use v4 for opaque identifiers; use v7 when you want time-ordered keys for database indexes.
Are these cryptographically secure?: Yes. Both v4 and v7 generation use crypto.getRandomValues() from the WebCrypto API, which draws from your operating system's CSPRNG. Math.random() is never used.
How likely is a collision?: For v4: practically zero. You would need to generate ~2.71 quintillion UUIDs to have a 50% chance of one collision. For v7 within the same millisecond, randomness covers the remaining 74 bits, so collisions in practice are also negligible at any normal volume.
Can I generate them in bulk?: Yes. Set the count and the tool will produce up to a million values at once, with options to copy them as a list, JSON array, CSV, or SQL VALUES clause.