Reviewing AI Like an Art Critic: Grok 4.6 Tested on Astrology & Philosophy

karinanguyen · x · 2026-08-13

Gaining early access to Grok 4.6, the author evaluated it alongside Grok 4.5, Claude Opus 5, and GPT-5.6-sol using an art critic's perspective. The evaluation used a 114-prompt set covering philosophy, self-knowledge, humor, poetry, and astrology. The author focused on weird model behaviors, personality quirks, and how they handle ambiguity, even bringing in a professional astrologer to independently assess the astrology-related answers.

Original post →

More from Models

Models channel →