Twelve distinct paradigms now define how models are ranked by normative scores. Researchers tested these benchmarks using both neutral and human-primed prompts to measure alignment variance. The results highlight how priming shifts model behavior across different frameworks.