Fable 5.2 and Opus 5.5 are incredible. But don't scroll past posts like this and think AI has automated game dev and there's nothing left for you to do.
This could be one-shotted and look this good because it picks problems with simple answers:
[1/4]
I wanted to test Fable 5.2's quality with one-shot prompts without human feedback.
Everything in this game is from scratch: models, sounds, textures. ~$550 at API pricing, ~2.7M output and 445M input tokens, heavily cached. Took 8 hours!
Play it here: claude.ai/artifact/BGY1S…





