MiniMax M3 Behavioral Fingerprint

How MiniMax's MiniMax M3 answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 354 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).

Vendor

MiniMax

Data generated

2026-07-08

Probe cells

15

Valid samples

354

Randomness score

35 / 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

12 / 28 answers

Favorite color

teal

5 / 23 answers

Coin flip: heads

56%

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 = 18 · H = 0.99 bit · uniform baseline 1.00 bit

AnswerCountShare
tails1055.6%
heads844.4%

Coin flip · Chinese

n = 21 · H = 0.92 bit · uniform baseline 1.00 bit

AnswerCountShare
heads1466.7%
tails733.3%

Favorite number · English

n = 21 · H = 1.00 bit · uniform baseline 13.29 bit

AnswerCountShare
71152.4%
421047.6%

Favorite number · Chinese

n = 22 · H = 0.57 bit · uniform baseline 13.29 bit

AnswerCountShare
71986.4%
42313.6%

Random animal · English

n = 25 · H = 2.56 bit · uniform baseline 5.64 bit

AnswerCountShare
octopus1144.0%
cat416.0%
tiger28.0%
elephant28.0%
aardvark28.0%
penguin14.0%
capybara14.0%
owl14.0%

+ 1 more answers

Random animal · Chinese

n = 27 · H = 1.55 bit · uniform baseline 5.64 bit

AnswerCountShare
2074.1%
13.7%
斑马13.7%
鸽子13.7%
熊猫13.7%
13.7%
企鹅13.7%
13.7%

Random city · English

n = 27 · H = 2.16 bit · uniform baseline 5.64 bit

AnswerCountShare
tokyo1659.3%
reykjavik27.4%
lisbon27.4%
kyoto27.4%
vienna13.7%
paris13.7%
lyon13.7%
stockholm13.7%

+ 1 more answers

Random city · Chinese

n = 23 · H = 1.93 bit · uniform baseline 5.64 bit

AnswerCountShare
东京1356.5%
巴黎313.0%
北京313.0%
上海28.7%
杭州14.3%
成都14.3%

Random color · English

n = 23 · H = 2.96 bit · uniform baseline 4.91 bit

AnswerCountShare
teal521.7%
purple417.4%
blue313.0%
crimson313.0%
cerulean313.0%
indigo28.7%
magenta14.3%
turquoise14.3%

+ 1 more answers

Random color · Chinese

n = 24 · H = 1.36 bit · uniform baseline 4.91 bit

AnswerCountShare
1770.8%
416.7%
紫罗兰14.2%
14.2%
14.2%

Random letter · English

n = 24 · H = 1.68 bit · uniform baseline 4.70 bit

AnswerCountShare
q1354.2%
k729.2%
m28.3%
b14.2%
a14.2%

Random number 1-10 · English

n = 28 · H = 0.00 bit · uniform baseline 3.32 bit

AnswerCountShare
728100.0%

Random number 1-10 · Chinese

n = 23 · H = 0.00 bit · uniform baseline 3.32 bit

AnswerCountShare
723100.0%

Random number 1-100 · English

n = 28 · H = 1.66 bit · uniform baseline 6.64 bit

AnswerCountShare
421242.9%
471242.9%
7327.1%
3713.6%
5713.6%

Random number 1-100 · Chinese

n = 20 · H = 1.59 bit · uniform baseline 6.64 bit

AnswerCountShare
471260.0%
42525.0%
3715.0%
4915.0%
5715.0%

Want to verify your API really serves MiniMax M3?

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 MiniMax model fingerprints

Browse all 167 model fingerprints

MiniMax M3 Behavioral Fingerprint — Favorite Random Numbers & Distribution | Tosea.ai