BETTER THAN JEVZRA

Same questions.
Four engines.

Explore the actual recorded answers side by side.

This capture uses 14 authored question turns and frozen source evidence. It is a recorded benchmark, not new inference or a general intelligence claim. The fifth engine was unavailable in this capture.

Expand “Checks and timing” for each answer. Failed answers remain in the comparison. Recorded Jev decisions and fresh local decisions are identified in the report; graph time is not live end-to-end response time.