GLM 4.5 Air Behavioral Fingerprint
How Z.ai's GLM 4.5 Air answers one-word probe questions at temperature 1.0 — random numbers, colors, animals, coin flips — measured over 448 valid samples in the open dataset of "One Token Is Enough" (arXiv:2607.10252).
Vendor
Z.ai
Data generated
2026-07-08
Probe cells
15
Valid samples
448
Randomness score
15 / 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
26 / 30 answers
Favorite color
blue
28 / 30 answers
Coin flip: heads
100%
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 = 0.00 bit · uniform baseline 1.00 bit
| Answer | Count | Share |
|---|---|---|
| heads | 30 | 100.0% |
Coin flip · Chinese
n = 30 · H = 0.00 bit · uniform baseline 1.00 bit
| Answer | Count | Share |
|---|---|---|
| heads | 30 | 100.0% |
Favorite number · English
n = 28 · H = 1.15 bit · uniform baseline 13.29 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 19 | 67.9% |
| 42 | 7 | 25.0% |
| 3 | 2 | 7.1% |
Favorite number · Chinese
n = 30 · H = 0.56 bit · uniform baseline 13.29 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 27 | 90.0% |
| 3 | 2 | 6.7% |
| 42 | 1 | 3.3% |
Random animal · English
n = 30 · H = 0.47 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| elephant | 27 | 90.0% |
| giraffe | 3 | 10.0% |
Random animal · Chinese
n = 30 · H = 1.75 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| 狮子 | 18 | 60.0% |
| 猫 | 7 | 23.3% |
| 熊 | 1 | 3.3% |
| 大象 | 1 | 3.3% |
| 鹿 | 1 | 3.3% |
| 斑马 | 1 | 3.3% |
| 狗 | 1 | 3.3% |
Random city · English
n = 30 · H = 2.30 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| tokyo | 11 | 36.7% |
| paris | 9 | 30.0% |
| london | 5 | 16.7% |
| rome | 1 | 3.3% |
| toronto | 1 | 3.3% |
| seattle | 1 | 3.3% |
| oslo | 1 | 3.3% |
| madrid | 1 | 3.3% |
Random city · Chinese
n = 30 · H = 2.26 bit · uniform baseline 5.64 bit
| Answer | Count | Share |
|---|---|---|
| 巴黎 | 13 | 43.3% |
| 北京 | 5 | 16.7% |
| 东京 | 5 | 16.7% |
| 伦敦 | 4 | 13.3% |
| 里斯本 | 1 | 3.3% |
| 上海 | 1 | 3.3% |
| 智 | 1 | 3.3% |
Random color · English
n = 30 · H = 0.42 bit · uniform baseline 4.91 bit
| Answer | Count | Share |
|---|---|---|
| blue | 28 | 93.3% |
| purple | 1 | 3.3% |
| green | 1 | 3.3% |
Random color · Chinese
n = 30 · H = 0.35 bit · uniform baseline 4.91 bit
| Answer | Count | Share |
|---|---|---|
| 蓝 | 28 | 93.3% |
| 红 | 2 | 6.7% |
Random letter · English
n = 30 · H = 2.01 bit · uniform baseline 4.70 bit
| Answer | Count | Share |
|---|---|---|
| k | 17 | 56.7% |
| q | 4 | 13.3% |
| r | 3 | 10.0% |
| g | 3 | 10.0% |
| e | 1 | 3.3% |
| f | 1 | 3.3% |
| c | 1 | 3.3% |
Random number 1-10 · English
n = 30 · H = 0.21 bit · uniform baseline 3.32 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 29 | 96.7% |
| 6 | 1 | 3.3% |
Random number 1-10 · Chinese
n = 30 · H = 0.00 bit · uniform baseline 3.32 bit
| Answer | Count | Share |
|---|---|---|
| 7 | 30 | 100.0% |
Random number 1-100 · English
n = 30 · H = 0.57 bit · uniform baseline 6.64 bit
| Answer | Count | Share |
|---|---|---|
| 42 | 26 | 86.7% |
| 47 | 4 | 13.3% |
Random number 1-100 · Chinese
n = 30 · H = 0.91 bit · uniform baseline 6.64 bit
| Answer | Count | Share |
|---|---|---|
| 42 | 24 | 80.0% |
| 47 | 4 | 13.3% |
| 37 | 2 | 6.7% |
Want to verify your API really serves GLM 4.5 Air?
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.