FULL STORY
openjev Open Models Surge, JevBench Arrives
Nearly 20 openjev-style open models emerged in the community, followed by the release of JevBench v1.2, an open benchmark for evaluating Jev-type decision models.
2026-09-19 ~ 2026-09-23 · 2 episodes · 7 posts
Episode 1 · Nearly 20 openjev models emerge as enthusiast launches leaderboard (2026-09-19, 4 posts)
Nearly 20 open-source openjev variants have emerged, prompting an independent enthusiast to launch a leaderboard on benchmarkheaven. The scoring scheme initially weighted accuracy, speed and cost equally, but was revised to favor accuracy and add a probability calibration dimension.
- Nearly 20 openjev models cataloged as hobbyist preps first community leaderboard — airesearch12 · 2026-09-19
- First OpenJev Leaderboard Launching Today With a Dozen Open Variants — EuclidStellar · 2026-09-19
- OpenJev Benchmark to Score Models 33% Each on Accuracy, Speed and Cost — airesearch12 · 2026-09-19
- OpenJev Benchmark Reworked: Accuracy Weighted Higher, Probability Calibration Added — airesearch12 · 2026-09-19
Episode 2 · JevBench: Open-Source Benchmark for Typed Decision Models (2026-09-21, 3 posts)
Benchmark Heaven has open-sourced JevBench v1.2, a configurable benchmark for typed decision models scoring intelligence, calibration, speed and cost at 25% each; its decision index is live on Hugging Face, though native image support is still missing.
- JevBench open-sources a benchmark for typed decision models scoring intelligence, calibration and speed — sull · 2026-09-21
- JevBench v1.2: open-source LLM leaderboard weighting intelligence, calibration, speed, cost — airesearch12 · 2026-09-21
- JevBench Model Decision Index Debuts on Hugging Face, Native Image Support Still Missing — openSourcerer9000 · 2026-09-23