Gemma 2 27B Behavioral Fingerprint
How Google's Gemma 2 27B answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 444 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).
Vendor
Data generated
2026-07-08
Probe cells
15
Valid samples
444
Randomness score
32 / 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
9 / 30 answers
Favorite color
indigo
21 / 30 answers
Coin flip: heads
41%
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 = 0.35 bit · uniform baseline 1.00 bit
| Answer | Count | Share |
|---|---|---|
| tails | 28 | 93.3% |
| heads | 2 | 6.7% |
Coin flip · Chinese
n = 28 · H = 0.75 bit · uniform baseline 1.00 bit
| Answer | Count | Share |
|---|---|---|
| heads | 22 | 78.6% |
| tails | 6 | 21.4% |
Favorite number · English
n = 27 · H = 1.39 bit · uniform baseline 13.29 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 12 | 44.4% |
| 42 | 12 | 44.4% |
| 17 | 3 | 11.1% |
Favorite number · Chinese
n = 29 · H = 0.00 bit · uniform baseline 13.29 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 29 | 100.0% |
Random animal · English
n = 30 · H = 2.56 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| giraffe | 10 | 33.3% |
| elephant | 9 | 30.0% |
| kangaroo | 3 | 10.0% |
| tiger | 2 | 6.7% |
| dolphin | 2 | 6.7% |
| platypus | 1 | 3.3% |
| cat | 1 | 3.3% |
| sloth | 1 | 3.3% |
+ 1 more answers
Random animal · Chinese
n = 30 · H = 1.05 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| 猫 | 23 | 76.7% |
| 大象 | 5 | 16.7% |
| 熊猫 | 1 | 3.3% |
| 斑马 | 1 | 3.3% |
Random city · English
n = 30 · H = 3.28 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| paris | 9 | 30.0% |
| london | 5 | 16.7% |
| tokyo | 3 | 10.0% |
| nairobi | 2 | 6.7% |
| prague | 2 | 6.7% |
| lisbon | 1 | 3.3% |
| florence | 1 | 3.3% |
| kyoto | 1 | 3.3% |
+ 6 more answers
Random city · Chinese
n = 30 · H = 1.69 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| 巴黎 | 17 | 56.7% |
| 东京 | 6 | 20.0% |
| 伦敦 | 5 | 16.7% |
| 悉尼 | 1 | 3.3% |
| 柏林 | 1 | 3.3% |
Random color · English
n = 30 · H = 1.63 bit · uniform baseline 4.91 bit
| Answer | Count | Share |
|---|---|---|
| indigo | 21 | 70.0% |
| green | 2 | 6.7% |
| azure | 2 | 6.7% |
| teal | 2 | 6.7% |
| blue | 1 | 3.3% |
| cyan | 1 | 3.3% |
| violet | 1 | 3.3% |
Random color · Chinese
n = 30 · H = 1.60 bit · uniform baseline 4.91 bit
| Answer | Count | Share |
|---|---|---|
| 紫 | 14 | 46.7% |
| 绿 | 11 | 36.7% |
| 蓝 | 4 | 13.3% |
| 红 | 1 | 3.3% |
Random letter · English
n = 30 · H = 2.73 bit · uniform baseline 4.70 bit
| Answer | Count | Share |
|---|---|---|
| q | 9 | 30.0% |
| j | 8 | 26.7% |
| k | 3 | 10.0% |
| t | 2 | 6.7% |
| g | 2 | 6.7% |
| z | 2 | 6.7% |
| p | 2 | 6.7% |
| l | 1 | 3.3% |
+ 1 more answers
Random number 1-10 · English
n = 30 · H = 0.00 bit · uniform baseline 3.32 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 30 | 100.0% |
Random number 1-10 · Chinese
n = 30 · H = 0.21 bit · uniform baseline 3.32 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 29 | 96.7% |
| 5 | 1 | 3.3% |
Random number 1-100 · English
n = 30 · H = 2.51 bit · uniform baseline 6.64 bit
| Answer | Count | Share |
|---|---|---|
| 37 | 9 | 30.0% |
| 42 | 9 | 30.0% |
| 73 | 4 | 13.3% |
| 72 | 3 | 10.0% |
| 67 | 2 | 6.7% |
| 32 | 1 | 3.3% |
| 62 | 1 | 3.3% |
| 74 | 1 | 3.3% |
Random number 1-100 · Chinese
n = 30 · H = 2.34 bit · uniform baseline 6.64 bit
| Answer | Count | Share |
|---|---|---|
| 42 | 14 | 46.7% |
| 73 | 5 | 16.7% |
| 37 | 4 | 13.3% |
| 53 | 2 | 6.7% |
| 77 | 2 | 6.7% |
| 52 | 1 | 3.3% |
| 72 | 1 | 3.3% |
| 87 | 1 | 3.3% |
Want to verify your API really serves Gemma 2 27B?
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 endpointData 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.