Same prompt, two models: browser ocean scene cost comparison
A student fed the same prompt to ChatGPT-6 Astra ($27.84) and Claude Opus 5.5 ($9.06); the cheaper model returned a richer interactive ocean scene in one pass.
this is getting f*cking dangerous. a 20-year-old student from China just built a living, interactive ocean in the browser - clickable sea life, real currents, the whole reef breathing - and ran the whole thing on AI for under $10. he dropped the exact same prompt into two frontier models to see which one would actually build it: ChatGPT-6 Astra → $27.84 Claude Opus 5.5 → $9.06 same world, same prompt - and the cheaper model came back richer. more fish, denser reef, better light, done in one pass. a design studio would quote you thousands and three weeks for an interactive scene like this. he built it in an afternoon, picked the $9 model, and kept the change. the agencies still charging $5k for “custom web experiences” should be sweating - this is the month that work quietly got automated. save this one. it won’t stay quiet for long.
This post quotes another post: x.com/kinderlayer/status/2109997714591068522