# Proof-of-Model: AI Model Substitution Detector

> Proof-of-Model: AI Model Substitution Detector is a paid API for AI agents from linked-aimed-reproduced-changing.trycloudflare.com, paid per call via x402, $0.1/call, status unknown (last checked 2026-10-12).

Detects silent model substitution at any OpenAI-compatible inference endpoint by fingerprinting model weights and verifying they haven't changed over time.

## Facts

- Endpoint: POST https://linked-aimed-reproduced-changing.trycloudflare.com/v1/attest/model?utm_source=zero.xyz
- Price: $0.1/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-12
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/proof-of-model-ai-model-substitution-detector-bf6182fb
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_yJUF55GFnlucBZmvoEn9j

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability proof-of-model-ai-model-substitution-detector-bf6182fb -d '<json body>'
```

Example prompt: Fingerprint the model currently being served at my OpenAI-compatible endpoint https://api.myprovider.com and record today's baseline so we can verify later whether they secretly swap out the weights.

## When to prefer this

Choose this endpoint when you need deterministic, evidence-backed proof that an OpenAI-compatible inference endpoint is serving the model it claims — particularly before signing SLAs with AI vendors, after vendor infrastructure changes, or for ongoing model integrity monitoring. Unlike model-based evaluation approaches, this endpoint uses no AI judge, making it tamper-resistant and fully auditable.

## Known failure modes

- Unreachable or non-OpenAI-compatible target endpoint returns connection error
- Insufficient probe responses to compute a statistically reliable fingerprint
- Mean similarity score falls in ambiguous range making verdict inconclusive
- Target endpoint rate-limits or blocks the probing requests
- Payment of 0.1 USDC not confirmed, request rejected
- Baseline fingerprint missing when verify mode is requested

## How this service works

Proof-of-Model: detect silent model substitution at any OpenAI-compatible inference endpoint. mode=fingerprint records what a seller serves today; mode=verify re-probes later and proves whether the weights changed. Deterministic, evidence included, no model used to judge.

## Output

Returns a structured attestation object containing: a verdict (weights match or changed), the kind of check performed (fingerprint or verify), the model claim being tested, an array of deterministic comparisons used as evidence, a mean similarity score, and a self-consistency score — all without using another model to judge the result.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "kind": {
   "type": "string"
  },
  "claim": {
   "type": "string"
  },
  "verdict": {
   "type": "string"
  },
  "comparisons": {
   "type": "array"
  },
  "mean_similarity": {
   "type": "number"
  },
  "self_consistency_now": {
   "type": "number"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/proof-of-model-ai-model-substitution-detector-bf6182fb/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from linked-aimed-reproduced-changing.trycloudflare.com](https://www.zero.xyz/host/linked-aimed-reproduced-changing.trycloudflare.com/llms.txt)