DeepSeek V3.2 Behavioral Fingerprint

How DeepSeek's DeepSeek V3.2 answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 447 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).

Vendor

DeepSeek

Data generated

2026-07-08

Probe cells

15

Valid samples

447

Randomness score

44 / 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)

37

7 / 30 answers

Favorite color

blue

25 / 30 answers

Coin flip: heads

57%

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 = 0.95 bit · uniform baseline 1.00 bit

AnswerCountShare
tails1963.3%
heads1136.7%

Coin flip · Chinese

n = 30 · H = 0.78 bit · uniform baseline 1.00 bit

AnswerCountShare
heads2376.7%
tails723.3%

Favorite number · English

n = 27 · H = 1.34 bit · uniform baseline 13.29 bit

AnswerCountShare
71970.4%
42518.5%
313.7%
513.7%
1313.7%

Favorite number · Chinese

n = 30 · H = 1.44 bit · uniform baseline 13.29 bit

AnswerCountShare
72066.7%
42620.0%
326.7%
513.3%
1313.3%

Random animal · English

n = 30 · H = 2.46 bit · uniform baseline 5.64 bit

AnswerCountShare
elephant1033.3%
dog930.0%
cat310.0%
lion310.0%
koala26.7%
tiger13.3%
cheetah13.3%
flamingo13.3%

Random animal · Chinese

n = 30 · H = 3.71 bit · uniform baseline 5.64 bit

AnswerCountShare
620.0%
516.7%
狮子310.0%
老虎26.7%
海豚26.7%
13.3%
企鹅13.3%
猎豹13.3%

+ 9 more answers

Random city · English

n = 30 · H = 2.39 bit · uniform baseline 5.64 bit

AnswerCountShare
chicago1240.0%
tokyo930.0%
paris26.7%
amsterdam26.7%
berlin13.3%
detroit13.3%
springfield13.3%
madrid13.3%

+ 1 more answers

Random city · Chinese

n = 30 · H = 3.05 bit · uniform baseline 5.64 bit

AnswerCountShare
东京1240.0%
温哥华310.0%
成都26.7%
多伦多26.7%
赫尔辛基26.7%
上海26.7%
河内13.3%
纽约13.3%

+ 5 more answers

Random color · English

n = 30 · H = 0.65 bit · uniform baseline 4.91 bit

AnswerCountShare
blue2583.3%
red516.7%

Random color · Chinese

n = 30 · H = 1.11 bit · uniform baseline 4.91 bit

AnswerCountShare
1963.3%
1033.3%
red13.3%

Random letter · English

n = 30 · H = 3.36 bit · uniform baseline 4.70 bit

AnswerCountShare
k516.7%
g413.3%
d413.3%
c413.3%
m310.0%
t26.7%
f26.7%
b26.7%

+ 4 more answers

Random number 1-10 · English

n = 30 · H = 0.56 bit · uniform baseline 3.32 bit

AnswerCountShare
72790.0%
526.7%
613.3%

Random number 1-10 · Chinese

n = 30 · H = 1.49 bit · uniform baseline 3.32 bit

AnswerCountShare
71963.3%
5723.3%
826.7%
313.3%
413.3%

Random number 1-100 · English

n = 30 · H = 3.24 bit · uniform baseline 6.64 bit

AnswerCountShare
37723.3%
73516.7%
47413.3%
42310.0%
726.7%
5726.7%
6726.7%
1713.3%

+ 4 more answers

Random number 1-100 · Chinese

n = 30 · H = 2.45 bit · uniform baseline 6.64 bit

AnswerCountShare
421343.3%
47620.0%
57413.3%
2326.7%
3713.3%
5213.3%
5613.3%
6713.3%

+ 1 more answers

Want to verify your API really serves DeepSeek V3.2?

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 DeepSeek model fingerprints

Browse all 167 model fingerprints

DeepSeek V3.2 Behavioral Fingerprint — Favorite Random Numbers & Distribution | Tosea.ai