Why is GPT-4-mini via Azure so much worse than Claude 3.7 for coding in CursorIDE?

Discussion
1

I have access to Azure through the Microsoft for Startups program.

I’m using GPT-4-mini via Azure for coding, but its response quality is worse than Claude 3.7—is that true?

I’ve been using Cursor with Claude for over a year, and switching to Azure yields poor results. The model seems weak for coding.

Is the problem with the GPT model itself or because I’m using an external Azure API instead of Cursor’s built-in subscription model?

2

Dude it’s the model

gpt 4 mini is a small model the code is ■■■■
3.7 is the advanced model nothing to do with cursor

3

I wouldn’t trust gpt-4o-mini to write ‘Hello World’
Use o4-mini instead if you have access.

It seems decent to me, but Gemini 2.5 Pro and Claude 3.7 are still king.