Jev Decision Index 0.4.1 - Full Score: leaderboard

Original Jev Decision Index0.4.1 release-v4, identical corpus b2b56d6fb636837ca469e689087bdbf373dda8de7638aa2da6793e6eda0792d5 in both pinned snapshots. Native notes explicitly distinguish submitted public0.3 runs adopted after kit rescore from private tests run by the benchmark lab on one RTX PRO6000 with each author serving. Agreement checks: Sieve100percent/831rows, GLiDE99.90percent/895rows against98.5percent criterion. Native exact Full score is0.60 private same-skills plus0.40 private new-domains, both mapped to public scale with untuned stock models; private denominators and mathematical endpoints are not stated, so no0–100 hard bound is invented. Sieve public/raw/breadth values are adopted mirrors of its separately ingested native release run and are not duplicated here.

Metric: Native Full index (private same-skills/new-domain stock-model mapped score). Source: huggingface.co. Saturation forecast: Rough model projection: around 2026. 2 models tracked. Updated 14 Nov 2026.

Top models

#ModelScoreOverall rank
1GLiDE v1.271.52
2Sieve-31B-v371.1

Interactive version: theaggregate.ai/benchmark?slug=jev-decision-index-0-4-1-full-score · How It Works · Data refreshed daily, snapshot 2026-11-16.