🧪 Test what your API endpoint actually delivers
Use one low-impact Quick Test to check whether an endpoint is reachable, whether the key is accepted, whether models can be discovered, and whether the selected model can complete both a chat and a streaming call. The result describes this one real observation only; it is not a permanent health check or proof of model identity.
The default probe sends few requests, no concurrency, no automatic retries. CORS or provider policy may block a direct browser connection; this version does not fall back to an APISpotlight proxy and does not scan third-party endpoints automatically.
THIRD-PARTY TOOL · NOT PROVIDED BY THIS SITE
Need a deeper authenticity check?
This page checks reachability, credentials, model discovery and the model identifier the response declares. It does not statistically fingerprint the model behind the endpoint. A separate third-party project does that by sampling answer distributions and comparing them against reference baselines.
meow llm detector (meow-llm-detector)
Ships GPT and Claude reference baselines and reports a three-colour verdict: a strong pointer to the declared model, insufficient evidence, or a strong pointer to another candidate model. Its author states explicitly that it is not a model identity card and cannot by itself prove that a vendor is substituting models.
- Online version (nothing to install):
- meowllm.top (spelled meow + llm) — public HTTPS APIs only. The author states that online-version requests, including your key, are processed by Cloudflare and the service's server, and that reports are public (request URL, group and sanitised errors; the key is not included).
- Local version:
- Download from GitHub Releases and run it on your own machine; it is the only option for private or intranet endpoints.
- License:
- PolyForm Noncommercial 1.0.0 — noncommercial use only. Personal research, experiments and testing are permitted; commercial use is not licensed by that project.
- Credit:
- Copyright 2026 chen-006 and contributors · Original project: github.com/chen-006/meow-llm-detector
Either way, use a separate, revocable key with a spending limit — the same advice the project's own README gives.
This site neither provides nor endorses that project, and does not run its code or its baseline data. Our own authenticity checks (scripts/authenticity_test.py) are an independent implementation, documented on the methodology page.