# AI use cases for Modelprint (unclecode/modelprint)

Repo: [unclecode/modelprint](https://github.com/unclecode/modelprint) | Use cases: 3 | Page: https://dedupedai.com/use-cases/unclecode-modelprint/

modelprint is a browser tool that runs infrastructure probes against OpenAI-compatible API endpoints and compares their fingerprints side by side.

This file holds every use case of this repo. Each one starts with a "## " heading and ends with its URL.

- Check which model a provider alias really serves: https://dedupedai.com/use-cases/unclecode-modelprint/verify-model-behind-api-alias/
- Run the test suites before opening a probe pull request: https://dedupedai.com/use-cases/unclecode-modelprint/check-probe-changes-before-pull-request/
- Rank a mystery model against a suspect lineup: https://dedupedai.com/use-cases/unclecode-modelprint/compare-stealth-model-against-suspect-lineup/

## How to check which model a provider alias really serves with Modelprint

Repo: [unclecode/modelprint](https://github.com/unclecode/modelprint) | Difficulty: Intermediate | Time: 15 minutes | Basis: From the README

Run the nine-probe browser suite to see if a named model matches its claimed lab.

**Who it is for:** An engineer or product lead who depends on a provider model name, such as one that stays the same while the served model changes behind it.

**The problem:** Providers can move models behind aliases and nothing tells you, so the model name on the invoice may not match what answers your calls.

**The outcome:** A side-by-side fingerprint table for the endpoint you test, which you can compare with the fingerprint of the model the provider claims to serve.

### Steps

1. Open https://modelprint.ai/ in your browser (the README calls it 'Try it: Modelprint.ai').
2. Enter the endpoint and your API key; the README says keys stay in your browser tab and calls go straight to the provider.
3. Select the model you want to check and run the probes.
4. Compare the tokenizer and error-code rows with the rows of the model the provider says it serves.
5. Treat matching rows as strong evidence about the model family, not proof of identity.

### Prompt to paste into your agent

```text
Using the modelprint repo (github.com/unclecode/modelprint), run the browser probe suite against <endpoint-or-model-name> and report the fingerprint table for each probe. Expected output: one value per probe, with tokenizer rows normalized against the one-character baseline. Verify that no tokenizer row is marked unstable before you call any match.
```

### Caveat

The README says some answers through a router are the router's, not the lab's, so pin a host where possible before trusting a row.

Risk, credentials: Your API key is entered in the browser page, so use a key you can rotate and do not run the page on a shared screen.

Dual-use notice: only use this on software, sites and systems you own or are authorized to test.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/unclecode-modelprint/verify-model-behind-api-alias/

## How to run the test suites before opening a probe pull request with Modelprint

Repo: [unclecode/modelprint](https://github.com/unclecode/modelprint) | Difficulty: Advanced | Time: 1 hour | Basis: From the README

Use selftest.mjs and smoke.mjs to confirm a new or edited probe behaves before review.

**Who it is for:** A contributor who has written or changed a probe in probes/ and wants to confirm it meets the project rules before opening a pull request.

**The problem:** A probe that throws, returns timestamps, or gives a non-deterministic value will not be accepted, and the README lists those rules.

**The outcome:** A local pass or fail result from the mock-harness suite, and printed values from two real models if you have a key.

### Steps

1. Clone the repository with `git clone https://github.com/unclecode/modelprint` and change into the folder. (inferred, check the README)
2. Copy `probes/_template.js` to a new file in `probes/` and write your probe following the contract in the README.
3. Run `node selftest.mjs` to test every probe against a mock harness; this needs no key.
4. Run `node smoke.mjs` with an OpenRouter key in the environment to print values from two real models.
5. Add your probe to `probes/index.js` and your name to CONTRIBUTORS.md, then open a pull request.

### Prompt to paste into your agent

```text
In the modelprint repo, add a probe in probes/ following probes/_template.js, then run `node selftest.mjs` from the repo root. Expected output: a pass result for every probe including the new one. Verify the new probe returns the same value on two runs and never throws before you open a pull request.
```

### Caveat

The README says approved probes are one line in probes/index.js and that the project has security and determinism gates, so a passing local run does not guarantee acceptance.

Dual-use notice: only use this on software, sites and systems you own or are authorized to test.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/unclecode-modelprint/check-probe-changes-before-pull-request/

## How to rank a mystery model against a suspect lineup with Modelprint

Repo: [unclecode/modelprint](https://github.com/unclecode/modelprint) | Difficulty: Advanced | Time: 30 minutes | Basis: From the README

Run suspects.mjs to rank a stealth model against a 12-model field by probe match.

**Who it is for:** A researcher or analyst who wants to know which lab's stack a new unnamed model might share, after it appears on a router.

**The problem:** Guessing the lab from marketing or a hot take gives weak evidence, and a single clue is easy to misread.

**The outcome:** A printed ranking of the full 12-model lineup, with the tokenizer match counts for each candidate.

### Steps

1. Clone the repository with `git clone https://github.com/unclecode/modelprint` and change into the folder. (inferred, check the README)
2. Set an OpenRouter API key in the environment that the script reads.
3. Run `node suspects.mjs` to test the full 12-model lineup.
4. Read the ranking and check the tokenizer counts, which the README uses as the main evidence.
5. Record the result as evidence about the model family, not a confirmed identity.

### Prompt to paste into your agent

```text
In the modelprint repo, run `node suspects.mjs` with the OpenRouter key set in the environment. Expected output: a ranked list of the 12 suspect models with probe match counts. Verify that the top candidate's tokenizer row matches on all four normalized counts before you report it as a family match.
```

### Caveat

The README warns that matching fingerprints prove shared infrastructure, not identity, because a lab can serve two different models on the same stack.

Risk, reverse engineering: This probes the serving infrastructure of a model to infer its lab, so keep the results as analysis and do not treat a match as proof of who runs it.

Dual-use notice: only use this on software, sites and systems you own or are authorized to test.

Disclaimer: Third-party repos we don't own or vouch for. AI-generated use cases can be wrong: review everything and run at your own risk. Full text: https://dedupedai.com/disclaimer/

URL: https://dedupedai.com/use-cases/unclecode-modelprint/compare-stealth-model-against-suspect-lineup/