benchmark source-backed
JevBench: an independent benchmark for typed decision models
A benchmark compares Jev and other typed decision systems using frozen tasks, explicit adapters and published result artifacts. Its score views combine capability, calibration, latency and estimated cost, with public and sealed evaluation material.
Notes
The repository publishes scoring implementations, frozen-method documents, aggregate results and per-task public evidence; the live board provides multiple score views for the measured systems.
source-backed — Public repository, docs or live artifact. About this label