Score and evaluate AV models
on real-world intelligence
APEX scores reaction time, anticipation, avoidance and confidence against real events your system has never seen, and reads how ready a given city is for it. Every definition is published before the run, so the score means the same thing to whoever reads it.
How can you use APEX?
There is still no driving test for an autonomous system, but APEX provides a readiness case a regulator can inspect.
Is your AI ready to drive
in the real world?
Compare your autonomous system directly against the human behaviors it aims to understand.
Quantify your stack’s reaction speed against average and high-performing human benchmarks in identical scenarios.
Measure predictive risk signals. Did the AV anticipate a hazard as effectively as a safe human driver?
Analyze danger-avoidance behaviors using our BADAS 2.0 model, trained on millions of near-miss events.
Achieve statistical significance without driving billions of physical miles, using our massive historical corpus.
Talk to us about scoring AV models.
Join the developers, insurers, and regulators setting the new standard.
Typically replies in 1–2 business days.