17,090+
Model verification runs
Completed endpoint verification checks counted in the public evidence layer.
Verify model identity before you pay
Verify LLM APIs and gateways before you trust them
Compare provider behavior against official model fingerprints, and check compatibility and latency before routing traffic to an LLM API or gateway.
- Your API key is only used for this run and never stored
- Compare against official fingerprints to detect mismatches
- Save your report code to reopen detailed results
Model verification activity
A live signal of how many third-party endpoints have been checked.
3,508+
Last 30 days
New model verification runs observed in the past month.
4,029+
Claim mismatches found
Endpoints that did not behave like the claimed model.
After the diagnosis
Not convinced by the result? Inspect the answers yourself.
Anonymous diagnosis provides a behavior signal, not a decision for you. Sign in to run transparent tasks—or your own real task—across 2–3 models and inspect the differences answer by answer.
Model Drift Monitor
Track whether official models behave differently over time across OpenAI, Azure, Anthropic, and Bedrock.
OpenAI and Azure GPT targets, plus Anthropic and Bedrock Claude targets, are tracked separately so provider routing changes do not pollute official-model baselines.
Open drift monitor →Claude Opus 5 / Anthropic
Claude Opus 5 / AWS Bedrock
GPT-5.6 Sol / Azure OpenAI
GPT-5.6 Sol / OpenAI
Why verify?
A model name is not enough
LLM APIs and gateways make integration easy, but they do not guarantee that an endpoint behaves like the model it claims to serve.
- Compare behavior against official model fingerprints
- Check OpenAI- or Anthropic-style API compatibility
- Measure latency before real usage
Recent verification activity
Recently completed endpoint verification runs.