JevBench v1.6.0 Retests 92 Models, Reshuffles Leaderboard and Adds 22 Languages

JevBench v1.6.0 retested all 92 Jev-class models on a fresh sealed set, with Quyet-1.0-Large topping the capability chart at 81.7. New per-language scoring across 22 languages shows 31B models holding strong while small 4B models collapse in multilingual capability.

2026-10-05 ~ 2026-10-05 · 3 related posts

• JevBench v1.6.0 re-tests 92 models from scratch, new No. 1 emerges — airesearch12 · 2026-10-05
• JevBench Adds 22-Language Breakdown, Small Models Collapse in Japanese and Danish — airesearch12 · 2026-10-05

1 near-duplicate retellings: airesearch12