OpenAI's models in my experience are just as good at coding now. But it may depend a lot on the language and task? I'm not sure. I haven't been writing a ton of code lately.