Krillion Statistics: What 77,100 Real Answers Reveal

Since launch day we have logged every Krillion daily answer sheet: 60 days (July 16 – September 13, 2026), 420 prompts, 77,100 accepted answers. This is the first complete statistical analysis of that archive — how the six rarity tiers are actually populated, how rare a perfect answer really is, and which answers players cannot stop typing first. Every number below is computed from the sheets we publish daily, and you can check every one of them yourself.

The inverted pyramid of Krillion rarity

Here is the whole game in one table — every accepted answer across 60 days, grouped by tier:

TierTotal answersAvg per prompt*Most in one prompt
Deep Cut56,640136.23,077
Rare11,87628.3338
Schooler5,80613.9126
Plankton1,7844.219
Too Clever5101.64
One in a Krillion4841.22

*Averaged over prompts where the tier appears: Deep Cut shows up in 416 of 420 prompts, Too Clever in only 318, and One in a Krillion in 410. Rare and Plankton appear in all 420.

The surprise is the shape. The roster is an inverted pyramid: Deep Cut, the second-highest tier, holds 73% of all accepted answers — 56,640 of 77,100 — while the bottom tier, Plankton, holds barely 2%. But the players form the opposite pyramid: the tiny Plankton list is where the crowd's answers concentrate, and the vast Deep Cut roster is where almost nobody lands. Rarity scoring inverts intuition — the biggest list is the hardest to hit, because each entry on it was found by almost no one. For what each tier feels like while playing, see the tier-by-tier explainer; this report is about the numbers underneath.

One in a Krillion: rarer than you think

The top tier's reputation undersells it. Across 420 prompts there are exactly 484 perfect answers — about 1.2 per prompt. Ten prompts have never produced one at all, and no prompt has ever had more than 2. A perfect answer is not a skill tier you climb into with practice alone; it is a ceiling that gets touched a couple of times a day across the entire player base. The full tier profile is in our One in a Krillion deep dive, and the strangest specimens — miracle fruit, the ophicleide, Jordan's 26 kilometers of coastline — are ranked in the hall of fame.

The biggest and smallest answer sheets on record

The average prompt accepts 183.6 answers, but the range is wilder than the average suggests:

  • Biggest: September 7's "name a Mario character" — 3,147 accepted answers, of which 3,077 were Deep Cuts. Four decades of games means every player knows a different Mario character, so the long tail runs for miles.
  • Smallest: August 18's "name a country that starts with the letter L" — just 9 accepted answers. Geography has a hard floor; there simply are not many L-countries, and the roster shows it.

The pattern generalizes: open-ended pop-culture prompts balloon into thousands of answers, while constrained factual prompts stay tiny — and the tiny ones are where a perfect answer is mathematically most within reach.

The Plankton hall of fame: America's first-instinct crown

Plankton is the tier for the answer everyone thought of. Some answers cannot stop sinking there. The most repeat-offending Plankton answers across 60 days:

  • United States — Plankton 14 times
  • United Kingdom — 10 times
  • Australia and France — 9 times each
  • Brazil, Canada, China, Russia — 8 times each

Yes — the undisputed king of first instincts is the United States. Whenever a prompt can be answered with a country, enough players type the obvious one to bury it at 10 points. If your first idea is a superpower, the data says: keep thinking.

What 77,100 answers tell you to do differently

Three takeaways, each straight from the table above:

  1. Skip the superpowers. Eight country names account for 74 Plankton landings. Any prompt answerable with a country rewards the second association, not the first.
  2. Aim for the fat part of the pyramid. Deep Cut holds 136 answers per prompt — it is the most populated tier and the most reachable good outcome. Hunting One in a Krillion (1.2 per prompt) is how players bust.
  3. Exploit constrained prompts. Tiny rosters like the L-countries sheet compress the whole game into a handful of slots — exactly where preparation beats luck.

The habit-level version of this is our strategy guide; drill it against real past sheets in practice mode, and check whether your own ideas would have counted with the answer checker.

Methodology and honest limits

Every figure here is computed from the game's publicly available daily data — the same sheets we republish, one page per day, in our answer archive (the pipeline is documented in how we source answers). You can recount any number in this article against those pages. Two honest caveats. First, we count accepted answers on the roster, not players or submissions — the game's submission counts are not public, so "United States was Plankton 14 times" means it made the most-common list 14 times, not that we know exactly how many people typed it. Second, the tier-per-prompt averages count only prompts where that tier appears, which is why Too Clever averages 1.6 over its 318 prompts rather than 1.2 over all 420. Where a choice like that changes a number, we've said so.

Frequently asked questions

How is Krillion rarity determined?

Every accepted answer is graded by how popular it was among that day's players: the commonest valid answers land in Plankton (10 points), and the rarest earn One in a Krillion (100). The game does not publish its exact thresholds, but our 60-day log of 77,100 answers shows how the six tiers behave in practice.

How many answers does a Krillion prompt have?

On average 183.6 accepted answers per prompt across the 420 prompts we have logged — but the spread is enormous. The biggest roster on record (name a Mario character, September 7, 2026) holds 3,147 answers; the smallest (name a country starting with L, August 18, 2026) holds just 9.

What is the rarest possible Krillion score?

A perfect 100 points — the One in a Krillion tier. Across 420 logged prompts there are only 484 such answers (about 1.2 per prompt), 10 prompts have never produced one at all, and no prompt has ever had more than 2.