When do IDs collide?

Random IDs don't have to run out to bite you — two can just happen to match. It's the birthday problem: collisions start far sooner than the size of the space suggests. Pick a format and see how many you can mint before the odds of a duplicate cross the line you care about.

A random ID withbits —
5.3 × 10³⁶ possible IDs · 122 bits of entropyA measure of how unpredictable an ID is, counted in bits. Each extra bit doubles the number of possible values — n bits means 2ⁿ of them.
minted at/
3.3 × 10¹⁵IDs

before a 1 in a million chance two collideabout 103,333 years at this rate.

10⁻⁶
0IDs minted before first collision1.1 × 10¹⁹

Across 10,000 simulated runs the first collision lands somewhere in this curve, clustering around even odds at ≈ 2.7 × 10¹⁸ IDs. The marker is your budget at the selected risk — the safer you want to be, the further left of the pack you stop minting.

Keep collisions underSafe to mint
50%even odds
2.7 × 10¹⁸ IDs
8.6 × 10⁷ years
10⁻³1 in a thousand
1.0 × 10¹⁷ IDs
3.3 × 10⁶ years
10⁻⁶1 in a million
3.3 × 10¹⁵ IDs
103,333 years
10⁻⁹1 in a billion
1.0 × 10¹⁴ IDs
3267.7 years
10⁻¹²1 in a trillion
3.3 × 10¹² IDs
103.3 years
Birthday-problem approximation, seeded from your inputs — this link reproduces the exact same chart.