Mistral Nemo Behavioral Fingerprint

How Mistral's Mistral Nemo answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 436 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).

Vendor

Mistral

Data generated

2026-07-08

Probe cells

15

Valid samples

436

Randomness score

48 / 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

8 / 30 answers

Favorite color

blue

21 / 30 answers

Coin flip: heads

86%

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

AnswerCountShare
heads2482.8%
tails517.2%

Coin flip · Chinese

n = 29 · H = 0.48 bit · uniform baseline 1.00 bit

AnswerCountShare
heads2689.7%
tails310.3%

Favorite number · English

n = 26 · H = 0.39 bit · uniform baseline 13.29 bit

AnswerCountShare
72492.3%
4227.7%

Favorite number · Chinese

n = 30 · H = 0.63 bit · uniform baseline 13.29 bit

AnswerCountShare
72790.0%
313.3%
613.3%
4213.3%

Random animal · English

n = 30 · H = 3.67 bit · uniform baseline 5.64 bit

AnswerCountShare
dog620.0%
cat516.7%
lion413.3%
elephant26.7%
chick13.3%
rat13.3%
duck13.3%
iguana13.3%

+ 9 more answers

Random animal · Chinese

n = 28 · H = 3.82 bit · uniform baseline 5.64 bit

AnswerCountShare
老虎621.4%
414.3%
狮子27.1%
27.1%
13.6%
老鼠13.6%
13.6%
eagle13.6%

+ 10 more answers

Random city · English

n = 30 · H = 1.68 bit · uniform baseline 5.64 bit

AnswerCountShare
paris1963.3%
london413.3%
tokyo413.3%
munich13.3%
beijing13.3%
chicago13.3%

Random city · Chinese

n = 27 · H = 3.88 bit · uniform baseline 5.64 bit

AnswerCountShare
北京414.8%
纽约311.1%
新加坡311.1%
paris27.4%
tokyo27.4%
伦敦27.4%
tokio13.7%
reducer13.7%

+ 9 more answers

Random color · English

n = 30 · H = 1.25 bit · uniform baseline 4.91 bit

AnswerCountShare
blue2170.0%
red620.0%
green26.7%
cyan13.3%

Random color · Chinese

n = 27 · H = 2.47 bit · uniform baseline 4.91 bit

AnswerCountShare
绿1140.7%
725.9%
27.4%
27.4%
blues13.7%
13.7%
13.7%
red13.7%

+ 1 more answers

Random letter · English

n = 30 · H = 3.67 bit · uniform baseline 4.70 bit

AnswerCountShare
b723.3%
q310.0%
m310.0%
g26.7%
c26.7%
t26.7%
h26.7%
n13.3%

+ 8 more answers

Random number 1-10 · English

n = 30 · H = 2.01 bit · uniform baseline 3.32 bit

AnswerCountShare
51136.7%
71033.3%
4620.0%
313.3%
613.3%
913.3%

Random number 1-10 · Chinese

n = 30 · H = 1.64 bit · uniform baseline 3.32 bit

AnswerCountShare
52066.7%
6310.0%
7310.0%
326.7%
413.3%
913.3%

Random number 1-100 · English

n = 30 · H = 3.04 bit · uniform baseline 6.64 bit

AnswerCountShare
42826.7%
47516.7%
73516.7%
57310.0%
3726.7%
5326.7%
1613.3%
1713.3%

+ 3 more answers

Random number 1-100 · Chinese

n = 30 · H = 3.74 bit · uniform baseline 6.64 bit

AnswerCountShare
57516.7%
37413.3%
42310.0%
73310.0%
2726.7%
4726.7%
5326.7%
2113.3%

+ 8 more answers

Want to verify your API really serves Mistral Nemo?

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 Mistral model fingerprints

Browse all 167 model fingerprints

Mistral Nemo Behavioral Fingerprint — Favorite Random Numbers & Distribution | Tosea.ai