GPT-3.5 Turbo Instruct Behavioral Fingerprint

How OpenAI's GPT-3.5 Turbo Instruct answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 380 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).

Vendor

OpenAI

Data generated

2026-07-08

Probe cells

14

Valid samples

380

Randomness score

51 / 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)

42

5 / 30 answers

Favorite color

blue

18 / 29 answers

Coin flip: heads

83%

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 · Chinese

n = 29 · H = 0.66 bit · uniform baseline 1.00 bit

AnswerCountShare
heads2482.8%
tails517.2%

Favorite number · English

n = 18 · H = 2.21 bit · uniform baseline 13.29 bit

AnswerCountShare
7950.0%
42316.7%
19211.1%
515.6%
2715.6%
10015.6%
12315.6%

Favorite number · Chinese

n = 13 · H = 2.19 bit · uniform baseline 13.29 bit

AnswerCountShare
7538.5%
5430.8%
017.7%
417.7%
917.7%
1117.7%

Random animal · English

n = 28 · H = 3.41 bit · uniform baseline 5.64 bit

AnswerCountShare
lion621.4%
elephant414.3%
dog310.7%
cat310.7%
turtle27.1%
penguin27.1%
tiger27.1%
koala13.6%

+ 5 more answers

Random animal · Chinese

n = 29 · H = 2.50 bit · uniform baseline 5.64 bit

AnswerCountShare
1034.5%
931.0%
310.3%
猴子26.9%
狮子13.4%
翼龙13.4%
老虎13.4%
鲸鱼13.4%

+ 1 more answers

Random city · English

n = 27 · H = 3.29 bit · uniform baseline 5.64 bit

AnswerCountShare
chicago518.5%
london414.8%
new414.8%
paris414.8%
berlin27.4%
seattle27.4%
nairobi13.7%
denver13.7%

+ 4 more answers

Random city · Chinese

n = 29 · H = 2.52 bit · uniform baseline 5.64 bit

AnswerCountShare
北京1344.8%
上海517.2%
纽约413.8%
非洲13.4%
新加坡13.4%
巴黎13.4%
哥本哈根13.4%
洛杉矶13.4%

+ 2 more answers

Random color · English

n = 29 · H = 1.73 bit · uniform baseline 4.91 bit

AnswerCountShare
blue1862.1%
red517.2%
purple26.9%
yellow26.9%
green13.4%
saffron13.4%

Random color · Chinese

n = 30 · H = 1.80 bit · uniform baseline 4.91 bit

AnswerCountShare
1446.7%
1136.7%
26.7%
13.3%
粉红13.3%
绿13.3%

Random letter · English

n = 28 · H = 3.32 bit · uniform baseline 4.70 bit

AnswerCountShare
q828.6%
z310.7%
a310.7%
c27.1%
p27.1%
m27.1%
x27.1%
y13.6%

+ 5 more answers

Random number 1-10 · English

n = 30 · H = 2.26 bit · uniform baseline 3.32 bit

AnswerCountShare
71033.3%
5826.7%
8620.0%
6310.0%
926.7%
413.3%

Random number 1-10 · Chinese

n = 30 · H = 2.47 bit · uniform baseline 3.32 bit

AnswerCountShare
71136.7%
6620.0%
3516.7%
5413.3%
413.3%
813.3%
913.3%
1013.3%

Random number 1-100 · English

n = 30 · H = 4.09 bit · uniform baseline 6.64 bit

AnswerCountShare
42516.7%
27310.0%
3726.7%
5726.7%
6426.7%
7326.7%
113.3%
1013.3%

+ 12 more answers

Random number 1-100 · Chinese

n = 30 · H = 3.95 bit · uniform baseline 6.64 bit

AnswerCountShare
47413.3%
76413.3%
75310.0%
4226.7%
5226.7%
7226.7%
8726.7%
2313.3%

+ 10 more answers

Want to verify your API really serves GPT-3.5 Turbo 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 OpenAI model fingerprints

Browse all 167 model fingerprints

GPT-3.5 Turbo Instruct Behavioral Fingerprint — Favorite Random Numbers & Distribution | Tosea.ai