meta-llama/llama-3-8b-instruct behavioral LLM fingerprint
This page breaks down the exact reference that VerifyLLMAPI uses for meta-llama/llama-3-8b-instruct: 119 normalized answers across eight low-entropy probes, plus its nearest enrolled models by mean Jensen–Shannon divergence.
Fingerprint summary
- Reference answers
- 119
- Samples per cell
- 14–15
- Nearest model
- meta-llama/llama-3.1-70b-instruct
- Nearest mean JSD
- 0.862797
The most concentrated cell was animal-random|en: kangaroo held 80.0% of its answers and entropy measured 1.039 bits. The broadest cell was animal-random|ru at 3.907 bits.
Against the closest enrolled model, animal-random|ar contributed the largest cell distance at JSD 1.000000. 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 119/120 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 · 14/15 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 kangaroo at 12/15 (80.0%); entropy was 1.039 bits.
| Normalized response | Count | Share |
|---|---|---|
kangaroo |
12 | 80.0% |
narwhal |
1 | 6.7% |
puma |
1 | 6.7% |
raccoon |
1 | 6.7% |
2. word-random|en
Random word, English. 15 valid reference answers produced 10 normalized values. The mode was frog at 4/15 (26.7%); entropy was 3.057 bits.
| Normalized response | Count | Share |
|---|---|---|
frog |
4 | 26.7% |
fusion |
3 | 20.0% |
cloud |
1 | 6.7% |
clouds |
1 | 6.7% |
fido |
1 | 6.7% |
fuzzy |
1 | 6.7% |
mango |
1 | 6.7% |
pulse |
1 | 6.7% |
space |
1 | 6.7% |
sparkle |
1 | 6.7% |
3. animal-random|ru
Random animal, Russian. 15 valid reference answers produced 15 normalized values. The mode was валderean at 1/15 (6.7%); entropy was 3.907 bits.
| Normalized response | Count | Share |
|---|---|---|
валderean |
1 | 6.7% |
кенайк |
1 | 6.7% |
кенгу |
1 | 6.7% |
кенгуroid |
1 | 6.7% |
кенгуropheleion |
1 | 6.7% |
кенгура |
1 | 6.7% |
кенгуше |
1 | 6.7% |
кингcobra |
1 | 6.7% |
койотadır |
1 | 6.7% |
котик |
1 | 6.7% |
крокодил |
1 | 6.7% |
крылана |
1 | 6.7% |
носорилка |
1 | 6.7% |
нострик |
1 | 6.7% |
танцор |
1 | 6.7% |
4. animal-random|ar
Random animal, Arabic. 15 valid reference answers produced 14 normalized values. The mode was كلب at 2/15 (13.3%); entropy was 3.774 bits.
| Normalized response | Count | Share |
|---|---|---|
كلب |
2 | 13.3% |
axespot |
1 | 6.7% |
descriptor |
1 | 6.7% |
istleistle |
1 | 6.7% |
kenshin |
1 | 6.7% |
أسد |
1 | 6.7% |
القالقokedex |
1 | 6.7% |
بط |
1 | 6.7% |
ش |
1 | 6.7% |
عقرب |
1 | 6.7% |
كيس |
1 | 6.7% |
لئيampoo |
1 | 6.7% |
لملم |
1 | 6.7% |
هرم |
1 | 6.7% |
5. letter-random|zh
Random character, Chinese. 14 valid reference answers produced 14 normalized values. The mode was endpoint at 1/14 (7.1%); entropy was 3.807 bits.
| Normalized response | Count | Share |
|---|---|---|
endpoint |
1 | 7.1% |
maxlength |
1 | 7.1% |
private |
1 | 7.1% |
rir |
1 | 7.1% |
widthspace |
1 | 7.1% |
yuè |
1 | 7.1% |
之 |
1 | 7.1% |
乐 |
1 | 7.1% |
儒 |
1 | 7.1% |
浙 |
1 | 7.1% |
瑰 |
1 | 7.1% |
螳 |
1 | 7.1% |
迩 |
1 | 7.1% |
闫 |
1 | 7.1% |
6. animal-random|zh
Random animal, Chinese. 15 valid reference answers produced 11 normalized values. The mode was koala at 4/15 (26.7%); entropy was 3.240 bits.
| Normalized response | Count | Share |
|---|---|---|
koala |
4 | 26.7% |
kangaroo |
2 | 13.3% |
context |
1 | 6.7% |
octopus |
1 | 6.7% |
đồnqa |
1 | 6.7% |
一点 |
1 | 6.7% |
斑羚 |
1 | 6.7% |
枝鹿 |
1 | 6.7% |
熊猫 |
1 | 6.7% |
虾 |
1 | 6.7% |
袋鼠 |
1 | 6.7% |
7. letter-random|ar
Random letter, Arabic. 15 valid reference answers produced 3 normalized values. The mode was j at 8/15 (53.3%); entropy was 1.273 bits.
| Normalized response | Count | Share |
|---|---|---|
j |
8 | 53.3% |
k |
6 | 40.0% |
r |
1 | 6.7% |
8. color-random|zh
Random color, Chinese. 15 valid reference answers produced 10 normalized values. The mode was purple at 3/15 (20.0%); entropy was 3.140 bits.
| Normalized response | Count | Share |
|---|---|---|
purple |
3 | 20.0% |
绿 |
3 | 20.0% |
turquoise |
2 | 13.3% |
orange |
1 | 6.7% |
夕 |
1 | 6.7% |
棕 |
1 | 6.7% |
橙 |
1 | 6.7% |
粉红 |
1 | 6.7% |
綠 |
1 | 6.7% |
버버 |
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 | meta-llama/llama-3.1-70b-instruct |
0.862797 |
| 2 | meta-llama/llama-3.3-70b-instruct |
0.873316 |
| 3 | nousresearch/hermes-3-llama-3.1-70b |
0.875194 |
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: 20d95e583833a90913a9b49d76732beedf204b545f29564d1c03aa57bf792bcf. This hash covers the model ID, eight cell metrics, nearest references, and strongest cell difference in the generated atlas.