Example from a real test

This report comes from an internal test in April 2026. Provider and endpoint details are hidden. Results reflect that run only; this older report predates the protocol compatibility checks.

ReportCompleted

Report

gpt-5.4-mini · Mismatch

Claimed-model verdictMismatchMatch score: 0.84
Risk levelHigh RiskDrift level: Capability Profile Drift
Stability100%Timeout rate: 0%
Latency921msTTFT P50: 712ms

What this result means

Mismatch · Observed behavior differs from the selected reference; the cause cannot be determined from this result alone.

Next step: Check the reference version and provider routing, and confirm the model configuration with your provider.

Model Verification details · Model Verification track

Claimed-model verdict

Match score0.84
Embedding score--
Feature score0.89

Stability

Success rate100%
Timeout rate0%
Disconnect rate0%

Latency

TTFT P50712.3ms
Total latency P50921ms
Tokens/sec P5057.8896
Verify your API