I recommend exploring different models using llama.cpp. It will recommend models that will work on your hardware, and you can download a coding-focused one from Hugging Face.