decisioncloud.OPEN INTELLIGENCE, SHARED.Get access

THE COMPARISON · SEPTEMBER 2026

Jev vs imajev.

Two decision models. Different ways to build.

The numbers.

JEVBENCH v1.4.2.2 · HIGHER IS BETTER
imajev-4B

67.37/ 100

JevBench composite score
Jev 1.13.0

63.29/ 100

JevBench composite score
JevBench component scores and sealed-set accuracy
Measureimajev-4BJev 1.13.0
Intelligence / 10052.253.1
Calibration / 10080.476.3
Sealed accuracy37.0%36.7%

imajev leads on the composite. Jev is slightly ahead on Intelligence. Neither wins every measure.

842 text decisions, including 308 sealed items. The composite combines intelligence, calibration, speed and cost; it is not accuracy. Self-hosted latency is adjusted and cost is estimated. These are external model results, not DecisionCloud API measurements.

Benchmark Heaven · JevBench v1.4.2.2 ↗Scored 27 Sep 2026 · Checked 29 Sep 2026 · Original data ↗

DECISIONCLOUD

Your next idea starts here.