SCOREBusiness workflows
Jev vs. Luna Performance Test
Jeff Fedor@a_jfedor
This project tested Jev against OpenAI's Luna on 20 synthetic supplier replies, evaluating performance and cost.
(1/3) Tested @typesafeai Jev vs @OpenAI Luna (reasoning off) on 20 synthetic supplier replies for Factory Twelve.
Same rubric, 140 labels
Results:
Jev: 139/140
Luna: 138/140
More importanlty:
Jev was 4.6× faster and ~83% cheaper at published rates.