Have you found GitHub copilot with Claude Sonnet 4.5 or GPT 5.2 to be the best for presenting the code in an easily read context?
Both are fine and add comments throughout their response. I'd use Claude over GPT any day though. Codex seems to do well on some tasks but 5.2 has made too many silly mistakes for me to use it again. Personally I stick with Opus but it has a x 3 multiplier in usage as compared to Sonnet and GPT models (consumes 3 x more in credits). But then if Sonnet and GPT take three - five times more attempts to complete something the cost/usage breaks even sooner or later as Opus one shots most things 99% of the time, and if it doesn't it is extremely complex (only failures I've had with Opus) and usually gets solved with a little prompt adjustment.
Welcome back!
Thanks Cheese!
Yeah, I’ve finally finished up in The Bahamas, now I’m retired and ready to start learning again. I’m very curious to learn more about how AI helps to sniff out errors and present remedies.
1 Like
ChatGPT 5.2 is doing well. However, I would prefer to work in the “sandbox” of Wappler if possible.