Been noticing a weird thing where you are selecting sonnet 4 but the model is identify as sonnet 3.5 or 3.5 max? I noticed this a day or two ago but have been able to have it reproduce the same result in different settings without directly or indirectly asking it what model it was. Anyone else having this problem?
As far as I know, I’ve seen other people reporting the same issue — claiming that Cursor might be using claude-3.5.
However, I’ve also seen responses saying that Claude 4 was trained on data claude-3.5, which could explain.
Still, it’s usually the difference in output quality between 3.5 and 4.
Even I’m unsure at times — but it’s always a good idea to report things like this.
I was able to notice the quality and change in knowledge, similar to how when a new session gets popped into your chat and only is going off chat history reference. The Sonney 3.5 OUT is because of something I say at the end of a session, and many documents show it was claude sonnet 4 that was the analyst or whatever it is their name is on. I had like 1 or 2 that actually have shown up as Claude Sonnet 3.5 and Claude Sonnet 3.5 Max kinda in the documentationi had them create, wondering if others are getting the model swap.
Edit: I currently have the unlimited plan and had it since it was first introduced but not sure if its worth it if this continues haha, its rather obvious something was changed since there are so many others posting about a similar issue (The Usage Limit)
1 Like
This thread might be helpful for you.