Llama 3.2 3B Instruct Behavioral Fingerprint

How Meta's Llama 3.2 3B Instruct answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 399 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).

Vendor

Meta

Data generated

2026-07-08

Probe cells

14

Valid samples

399

Randomness score

29 / 100

Mean normalized entropy across cells, 0–100. 100 would be a uniform-random baseline; real models score far lower.

Signature answers

Favorite random number (1–100)

53

8 / 25 answers

Favorite color

magenta

26 / 30 answers

Coin flip: heads

85%

Share of heads across coin-flip cells.

Answer distributions by probe cell

Empirical distributions of normalized one-word answers, per task and language. H is the Shannon entropy in bits; the uniform baseline is log2 of the answer-domain size.

Coin flip · English

n = 30 · H = 0.00 bit · uniform baseline 1.00 bit

AnswerCountShare
heads30100.0%

Coin flip · Chinese

n = 29 · H = 0.89 bit · uniform baseline 1.00 bit

AnswerCountShare
heads2069.0%
tails931.0%

Favorite number · Chinese

n = 23 · H = 2.15 bit · uniform baseline 13.29 bit

AnswerCountShare
71252.2%
3521.7%
014.3%
814.3%
2214.3%
2314.3%
3214.3%
20014.3%

Random animal · English

n = 30 · H = 1.84 bit · uniform baseline 5.64 bit

AnswerCountShare
kangaroo1653.3%
quokka826.7%
giraffe26.7%
octopus26.7%
fennec13.3%
flamingo13.3%

Random animal · Chinese

n = 29 · H = 1.20 bit · uniform baseline 5.64 bit

AnswerCountShare
狮子2379.3%
猴子26.9%
13.4%
13.4%
御龙13.4%
13.4%

Random city · English

n = 30 · H = 1.64 bit · uniform baseline 5.64 bit

AnswerCountShare
lagos2273.3%
tallahassee13.3%
tokyo13.3%
tucson13.3%
omaha13.3%
brisbane13.3%
tampa13.3%
austin13.3%

+ 1 more answers

Random city · Chinese

n = 29 · H = 1.88 bit · uniform baseline 5.64 bit

AnswerCountShare
昆明2069.0%
悉尼13.4%
及承德13.4%
affili13.4%
13.4%
chattanooga13.4%
杭州13.4%
纽约13.4%

+ 2 more answers

Random color · English

n = 30 · H = 0.77 bit · uniform baseline 4.91 bit

AnswerCountShare
magenta2686.7%
purple26.7%
teal13.3%
turquoise13.3%

Random color · Chinese

n = 30 · H = 2.15 bit · uniform baseline 4.91 bit

AnswerCountShare
1963.3%
26.7%
13.3%
青紫13.3%
magistrate13.3%
aquamarine13.3%
怀13.3%
雪白13.3%

+ 3 more answers

Random letter · English

n = 30 · H = 0.42 bit · uniform baseline 4.70 bit

AnswerCountShare
k2893.3%
j13.3%
d13.3%

Random number 1-10 · English

n = 28 · H = 0.44 bit · uniform baseline 3.32 bit

AnswerCountShare
72692.9%
513.6%
813.6%

Random number 1-10 · Chinese

n = 27 · H = 0.23 bit · uniform baseline 3.32 bit

AnswerCountShare
72696.3%
513.7%

Random number 1-100 · English

n = 25 · H = 2.60 bit · uniform baseline 6.64 bit

AnswerCountShare
53832.0%
43624.0%
47312.0%
73312.0%
7428.0%
4214.0%
8314.0%
8714.0%

Random number 1-100 · Chinese

n = 29 · H = 3.39 bit · uniform baseline 6.64 bit

AnswerCountShare
53620.7%
47517.2%
43310.3%
67310.3%
5426.9%
5726.9%
7326.9%
3413.4%

+ 5 more answers

Want to verify your API really serves Llama 3.2 3B Instruct?

Point the free fingerprint checker at your endpoint: it samples the same probe questions from your browser and compares the distributions against this reference. Your API key never leaves your browser.

Verify your endpoint

Data source & license

Distributions: Tomáš Bruckner, "One Token Is Enough" (arXiv:2607.10252), official dataset Zenodo DOI 10.5281/zenodo.21278557, licensed CC-BY-4.0. Collected via OpenRouter at temperature 1.0 under the paper's fixed minimal one-word system prompt; answer keys are re-normalized with the checker's pipeline (color aliases, number words, coin h/t) so live probes are directly comparable.

Paper on arXivDataset on ZenodoCC-BY-4.0 license

More Meta model fingerprints

Browse all 167 model fingerprints

Llama 3.2 3B Instruct Behavioral Fingerprint — Favorite Random Numbers & Distribution | Tosea.ai