Image JevBench v0.3.0 · individual system
djev-distill-v4 (tarsur385)
distilled · 26B-A4BBase model: google/diffusiongemma-26B-A4B-itsourcelocal/GPU evaluation
Image JevBench v0.3.0 composite score
3.616
Rank #32 of 44 ranked systems.
From the published full-benchmark aggregate. The composite combines intelligence, calibration, speed and cost.
Published axes
- intelligence
- 18.8
- calibration
- 61.7
- speed
- 82.9
- cost
- 41.1
Cost and speed evidence
- Cost per 1,000 decisions
- USD 0.071760
- Cost basis
- measured GPU seconds x $1.19/GPU-hour
- Measured latency · p50 / p95
- 0.228 s / 0.350 s
- Adjusted latency · p50 / p95
- 0.606 s / 0.851 s
- Latency adjustment
- v1.4 local latency adjustment
- Public accuracy
- 44.7%
- Sealed accuracy
- 45.6%
Published 2026-10-06. Scores and ranks can change in a later release. See the method notes.