Don’t take our word for it.
Run the MAE.
Choose a repeatable dataset, inspect the ground truth and observations, then compare estimation error. The proprietary Lexium estimator remains behind a local black-box boundary.
01 / CHOOSE DATASET
Curated evaluation datasets
02 / RUN THE MAE
Sudden state change
Tests recovery after a +22 state transition at sample 80.
Public mode exposes datasets and evidence only. The proprietary runtime is not shipped to the browser.
TruthObservationKalmanLexium
03 / FROZEN EVIDENCE
Recorded benchmark snapshot
This is a separately frozen benchmark record, not the selected browser dataset above.
Synthetic frozen benchmark only; not a universal real-world superiority claim.
Dataset recipes, observations, truth, benchmark metrics and exported results.
WHAT IS NOT EXPOSEDLexium source, internal state logic, proprietary formulas, private repositories or development assets.