knowing which model to use is more important than the prompt

o3 is best when giving codebase context and acting as an architect/planner

claude 4 opus is absolutely ripping defined PRDs

gemini 2.5 is best for specific small tasks

gpt 4.5 is my favorite writer, gemini a close second

o3/4o is the best image generator

I just realized GitHub bundles all these- is that kind of the typical setup right now to use a bundle package?

I’m not a professional full stack dev but I like working setups that are familiar to those that are at least. Especially if it saves me from having more than 1 AI subscription service for editing and pre-research context crawling.