6h agoGreg Kamradt finds Claude Opus 4.8 shows higher reasoning abstraction than Opus 4.7 on ARC-AGI-3 reasoning logsAn automated AmpCode judge evaluated the model reasoning logs.SentimentSentimentPos50%Neg50%Positive users praise Claude Opus 4.8 for its knowledge gains and rapid ARC-AGI-3 progress, while negative users call the low scores misleading and overpriced.12 comments with sentiment. View comments.