FINAL-Bench/Darwin-397B-ZTC
Text Generation • 404B • Updated • 1.1k • 33
ZTC Models: JEV ecosystems
Answer verification from one forward pass, zero generated tokens. Measured against JEV and 10 others on 2,018 identical items.
-
-
FINAL-Bench/ZTC-Judge-27B
Text Classification • 28B • Updated • 367 • 40
FINAL-Bench/ZTC-Judge-9B
Text Classification • 10B • Updated • 2.37k • 32Note Size ladder rung. Leaderboard AUC 0.6506 — above the surface baseline overall, and clearly so on professional exams.
FINAL-Bench/ZTC-Judge-4B
Text Classification • 5B • Updated • 125 • 33Note Size ladder rung. Leaderboard AUC 0.6360 — above the surface baseline overall, and clearly so on professional exams.
-
Gate Arcade
🚦33Plain vs JEV vs ZTC: a gate decides execute or hold
Typed Decision Leaderboard
🎯36JEV & open-Jev answer verifiers, one identical test set
Note Independent leaderboard — 13 answer verifiers on 2,018 identical items. ZTC (397B) 0.7364 leads; the gap to second place is not statistically distinguishable.
Verifier Playground
🧪32Run ZTC, JEV and Laya on the same question and answer
Note Paste a question and an answer; ZTC, JEV and Laya score it side by side on the same input.