Muse models rank friendliest, while GPT-6 and Grok lead on directness in AA-PersonalityBench

ArtificialAnlys says AA-PersonalityBench measures friendliness as warmth and approachability, directness as getting to the point, and Answer First as leading with the answer. Muse Spark 1.3 scores 100 for friendliness, GPT-6 Sol scores 100 for directness, and Grok 4.5 scores 100 for Answer First. Grok 4.6 scores 98 for directness but 8 for friendliness.