🩺 AX Diagnostic Leaderboard

Diagnose LLMs like a clinical exam — then publish the scores. Model quality and serving level, measured.

Evaluation axes (7)

Leaderboard — AX Index

#	Model	AX Index	Grade	trap_rate ↓	Health	tok/s ↑

◦ Seed data (metacognition trap_rate & serving from live measurements; other axes populate as diagnostics run).

Operated by FINAL-Bench (neutral benchmark operator) · Aligned with HELM · TrustLLM · MLPerf · NIST AI RMF · Korea AI Framework Act · Results public, diagnostic methods proprietary.