Spent the day with https://huggingface.co/lmstudio-community/Qwen2.5-Coder-14B-Instruct-MLX-8bit

Running it with #lmstudio on an M4 pro, performance is very good, results of coding prompts to generate #csharp or #golang have been spot on with very few exceptions.

If you're a dev in any language, you need to wrap your head around this.