What model does the Cursor team recommend?

General
1

With all the options, what model do you recommend of the most capable choices (gpt-4o, gpt-4, opus, etc)? Does it depend on the codebase, and if so, what are the considerations?

(And a feature suggestion - mark one of them as recommended in the UI so people know, especially as it may change!)

1 Like
2

I’m also curious about this. I’m not sure which model will give the best results. GPT-4 seems to have the annoying tendency to be “lazy” and only write a tiny bit of the code I’d like it to write.

3

Right now, I’d recommend gpt-4! I’ve been a bit surprised by gpt-4o’s performance on code.

2 Likes