Hands-On with Claude Fable 5.2: From Coding to Modeling and Video Creation—It’s Taking on Creative Work, Too

As AI continues to shatter records across benchmarks in reasoning, coding, and mathematics—making such breakthroughs feel almost routine—Anthropic’s late-night release of Claude Fable 5.2 redefined expectations in a far more direct and visceral way: it demonstrated a quantum leap in aesthetic perception, originality, and mastery of detail. With just a single prompt, the model autonomously produces an end-to-end video—from research and visual design to original music composition and video editing. To faithfully reconstruct the architectural space of Apple’s Jing’an Store in Shanghai, it spent over two hours analyzing hundreds of reference images and programmatically rebuilt the 3D structure. Crucially, its autonomous judgment has markedly improved: it proactively retrieves information, weighs design alternatives, rejects technically infeasible approaches, and even applies intrinsic quality standards to evaluate the final output. This marks a pivotal shift—from large models passively executing instructions toward genuine, independent creation. When AI…