GPT-5.6 Sol, Claude Opus 4.8, and Grok 4.5 compared on 100 frontend briefs

A Reddit user tested three frontier models on 100 frontend design briefs, finding each model tends to repeat similar visual patterns. The comparison includes 300 generated websites for analysis.