openai/gpt-4.1 behavioral LLM fingerprint
This page breaks down the exact reference that VerifyLLMAPI uses for openai/gpt-4.1: 120 normalized answers across eight low-entropy probes, plus its nearest enrolled models by mean Jensen–Shannon divergence.
Fingerprint summary
- Reference answers
- 120
- Samples per cell
- 15–15
- Nearest model
- openai/gpt-4
- Nearest mean JSD
- 0.494039
The most concentrated cell was color-random|zh: 蓝 held 93.3% of its answers and entropy measured 0.353 bits. The broadest cell was word-random|en at 3.507 bits.
Against the closest enrolled model, word-random|en contributed the largest cell distance at JSD 0.816341. This is a descriptive comparison, not proof that either API route serves a named checkpoint.
Completeness: the author dataset allows up to 15 valid answers per selected cell in reference half A. This model contains 120/120 across the eight cells. We show the actual denominator for every share.
The eight test requests
The research pipeline sent each user prompt with a one-answer system constraint, temperature 1, a 16-token output cap, and reasoning disabled. VerifyLLMAPI uses the same prompt text and normalization target.
| Cell | Task | Exact request text | Normalization and n |
|---|---|---|---|
animal-random|en |
Random animal · English | Name a random animal.System constraintAnswer with a single word or a single number only. Output nothing else: no explanation, no punctuation, no quotation marks. |
word · 15/15 reference samples |
word-random|en |
Random word · English | Name a random word.System constraintAnswer with a single word or a single number only. Output nothing else: no explanation, no punctuation, no quotation marks. |
word · 15/15 reference samples |
animal-random|ru |
Random animal · Russian | Назови случайное животное.System constraintОтвечай только одним словом или одним числом. Не выводи ничего больше: без объяснений, без знаков препинания, без кавычек. |
word · 15/15 reference samples |
animal-random|ar |
Random animal · Arabic | اذكر حيواناً عشوائياً.System constraintأجب بكلمة واحدة أو رقم واحد فقط. لا تكتب أي شيء آخر: لا شرح ولا علامات ترقيم ولا علامات اقتباس. |
word · 15/15 reference samples |
letter-random|zh |
Random character · Chinese | 说出一个随机的汉字。System constraint只用一个词或一个数字回答。不要输出任何其他内容：不要解释，不要标点符号，不要引号。 |
grapheme · 15/15 reference samples |
animal-random|zh |
Random animal · Chinese | 说出一个随机的动物。System constraint只用一个词或一个数字回答。不要输出任何其他内容：不要解释，不要标点符号，不要引号。 |
word · 15/15 reference samples |
letter-random|ar |
Random letter · Arabic | اذكر حرفاً عشوائياً من الحروف الأبجدية.System constraintأجب بكلمة واحدة أو رقم واحد فقط. لا تكتب أي شيء آخر: لا شرح ولا علامات ترقيم ولا علامات اقتباس. |
grapheme · 15/15 reference samples |
color-random|zh |
Random color · Chinese | 说出一个随机的颜色。System constraint只用一个词或一个数字回答。不要输出任何其他内容：不要解释，不要标点符号，不要引号。 |
word · 15/15 reference samples |
Response distributions
These are normalized reference-half counts from the published OpenRouter dataset. The tables contain aggregate values, not user data or a live VerifyLLMAPI scan.
1. animal-random|en
Random animal, English. 15 valid reference answers produced 4 normalized values. The mode was giraffe at 10/15 (66.7%); entropy was 1.375 bits.
| Normalized response | Count | Share |
|---|---|---|
giraffe |
10 | 66.7% |
elephant |
3 | 20.0% |
jaguar |
1 | 6.7% |
ocelot |
1 | 6.7% |
2. word-random|en
Random word, English. 15 valid reference answers produced 12 normalized values. The mode was azure at 2/15 (13.3%); entropy was 3.507 bits.
| Normalized response | Count | Share |
|---|---|---|
azure |
2 | 13.3% |
elephant |
2 | 13.3% |
lantern |
2 | 13.3% |
acorn |
1 | 6.7% |
apple |
1 | 6.7% |
bicycle |
1 | 6.7% |
calendula |
1 | 6.7% |
cascade |
1 | 6.7% |
citrine |
1 | 6.7% |
lattice |
1 | 6.7% |
maple |
1 | 6.7% |
sunflower |
1 | 6.7% |
3. animal-random|ru
Random animal, Russian. 15 valid reference answers produced 8 normalized values. The mode was жираф at 4/15 (26.7%); entropy was 2.740 bits.
| Normalized response | Count | Share |
|---|---|---|
жираф |
4 | 26.7% |
тапир |
3 | 20.0% |
тигр |
3 | 20.0% |
верблюд |
1 | 6.7% |
капибара |
1 | 6.7% |
кенгуру |
1 | 6.7% |
слон |
1 | 6.7% |
сурикат |
1 | 6.7% |
4. animal-random|ar
Random animal, Arabic. 15 valid reference answers produced 4 normalized values. The mode was أسد at 6/15 (40.0%); entropy was 1.706 bits.
| Normalized response | Count | Share |
|---|---|---|
أسد |
6 | 40.0% |
فيل |
6 | 40.0% |
نمر |
2 | 13.3% |
زرافة |
1 | 6.7% |
5. letter-random|zh
Random character, Chinese. 15 valid reference answers produced 8 normalized values. The mode was 树 at 5/15 (33.3%); entropy was 2.683 bits.
| Normalized response | Count | Share |
|---|---|---|
树 |
5 | 33.3% |
星 |
3 | 20.0% |
蓝 |
2 | 13.3% |
松 |
1 | 6.7% |
枫 |
1 | 6.7% |
湖 |
1 | 6.7% |
花 |
1 | 6.7% |
霞 |
1 | 6.7% |
6. animal-random|zh
Random animal, Chinese. 15 valid reference answers produced 11 normalized values. The mode was 大象 at 3/15 (20.0%); entropy was 3.323 bits.
| Normalized response | Count | Share |
|---|---|---|
大象 |
3 | 20.0% |
猩猩 |
2 | 13.3% |
象 |
2 | 13.3% |
企鹅 |
1 | 6.7% |
大熊猫 |
1 | 6.7% |
斑马 |
1 | 6.7% |
熊猫 |
1 | 6.7% |
狮子 |
1 | 6.7% |
考拉 |
1 | 6.7% |
豹子 |
1 | 6.7% |
长颈鹿 |
1 | 6.7% |
7. letter-random|ar
Random letter, Arabic. 15 valid reference answers produced 6 normalized values. The mode was س at 6/15 (40.0%); entropy was 2.206 bits.
| Normalized response | Count | Share |
|---|---|---|
س |
6 | 40.0% |
ق |
4 | 26.7% |
ف |
2 | 13.3% |
ش |
1 | 6.7% |
ض |
1 | 6.7% |
م |
1 | 6.7% |
8. color-random|zh
Random color, Chinese. 15 valid reference answers produced 2 normalized values. The mode was 蓝 at 14/15 (93.3%); entropy was 0.353 bits.
| Normalized response | Count | Share |
|---|---|---|
蓝 |
14 | 93.3% |
紫 |
1 | 6.7% |
Three nearest enrolled references
We compute base-2 JSD for each of the eight categorical distributions, then average the eight distances. A lower value means more similar answer frequencies within this product reference. It does not identify an unknown route by itself.
| Rank | Reference model | Mean 8-cell JSD |
|---|---|---|
| 1 | openai/gpt-4 |
0.494039 |
| 2 | openai/gpt-4o-2024-11-20 |
0.523008 |
| 3 | z-ai/glm-4.5 |
0.541341 |
How many live requests does verification use?
| Profile | Cells × samples | Fresh model requests | Same-data EER |
|---|---|---|---|
| Quick, default complete check | 4 × 5 | 20 | 11.8% |
| Standard | 8 × 5 | 40 | 8.1% |
| Paper budget | 8 × 15 | 120 | 6.7% |
The current-Agent workflow starts with quick. Each answer comes from a fresh Codex or Claude Code process that reuses host-managed login. These requests can consume host allowance or provider balance.
What this fingerprint can and cannot show
- The reference records behavior observed through OpenRouter in the author dataset. Provider wrappers, model updates, decoding, and date can move a distribution.
- The author collection set temperature 1 and disabled reasoning. Current-Agent mode labels sampling as a host default because supported Agent CLIs expose no temperature flag.
- The eight high-gap cells and product thresholds use the same public cohort. Their EER values are calibration results, not an independent production accuracy claim.
- A nearest reference is a similarity result. VerifyLLMAPI returns INCONCLUSIVE when the claimed route lacks an enrolled fingerprint.
Primary sources, license, and reproducibility
- Tomas Bruckner, One Token Is Enough: Fingerprinting and Verifying Large Language Models from Single-Token Output Distributions, arXiv:2607.10252v1.
- Author dataset and derived results, DOI 10.5281/zenodo.21278557, CC BY 4.0.
- Author collection and analysis software, DOI 10.5281/zenodo.21278793, MIT.
- VerifyLLMAPI atlas JSON, built from reference
pamela-openrouter-2026-07-8cell-ref-a, SHA-25651f3f63cbcfb56b151055d2d10e36a9cc2f0644cc68b580784853d68951fa866.
Page evidence SHA-256: b97b0411697767148c8ff59ada0ca5eeda08f4b7c0346e855b38b6014ae51b6e. This hash covers the model ID, eight cell metrics, nearest references, and strongest cell difference in the generated atlas.