Okay, I'll delve deeper. You're talking about one thing, the models, which Cursor has a bunch available to choose from. Yes, I agree practically no difference there in terms of what junk is being spit out.
But code review is really important to me and nothing comes close to Cursor in regards to reviewing the code the LLM generates. I can keep the good parts, modify or throw out the bad parts. I can go back to a checkpoint easily when things get really bad. What solution comes even close to that? Cursor nails this really well. Claude code acolytes tell me to just use git commands. Yeah, no thanks.
Cursor’s killer feature is that you can use it to edit diffs or restore a known good version? That is basic version control functionality. I understand what you’re referring to (I use Zed, which also has an interface for partially applying AI-generated code changes), but it’s very weird to me that this basic functionality would be considered some kind of competitive advantage.
you lost me here because this is based on your opinion and impressions, not data. How do you know nothing comes close?
The PR experience you describe is available in several options and setups. Different strokes for different folks, your choice is not superior by any meaningful measure other than your own preferences
But code review is really important to me and nothing comes close to Cursor in regards to reviewing the code the LLM generates. I can keep the good parts, modify or throw out the bad parts. I can go back to a checkpoint easily when things get really bad. What solution comes even close to that? Cursor nails this really well. Claude code acolytes tell me to just use git commands. Yeah, no thanks.
Vibe coding? Yes, no meaningful difference.