yep - its actually great for my use case /shrug
i dont prompt, i use orchestration, 5’s higher context window useful for that. using the custom GPT with 5 oh man, I can tell it to code for me and it follows instructions better than 4.
i imagine they did this to accomplish 2 things, making the model affordable without gimping devs who use local orchestration.
gate the model for being too lenient on the gates required to develop.
in either scenario 5 has been great for me! kudos