I may just be bad at writing coding prompts properly but recently I’ve felt like ChatGPT models try to intentionally maximize the time spent getting a coding answer to what you actually want as compared to Claude sonnet 4.5.
For example it would take me 2-3 prompts to debug a piece of code with the right solution in comparison to Claude getting that same solution in the first try with the same first prompt.
