I completely understand the context limiting. You all want to be a profitable business and probably were bleeding a ton of cash on model spend.
I’d like to have an account where I just pay you to upkeep cursor, but it comes with 0 model usage. I basically have to provide my own api keys. I realize that requests go through your backend and I would be paying for that as well.
Thoughts?
Hey, this is basically what the Pro plan is!
You can already put API keys in to use models at cost, but I don’t forsee a world where we offer an editor-only plan for any cheaper, as every request you make in Cursor always touches our own custom models in some way, even if you use an API key for the big model that responds!
I didn’t think custom models came into play for regular usage.
Well.. no matter. One other question.
I’ve seen people post saying that non max mode code now gets truncated. I see here Cursor – Models that ( for example ) the context window for claude 3.7 ( non max ) is 120k. Does this mean that truncating starts happening at 120k? If I get the message saying it would be best to start a new thread does that mean I’m above 120k? ( and not using max )
For reference here is an article that has me a bit concerned. Always Use Max Models in Cursor IDE - Vincent Schmalbach. But if he is just wrong and summation and/or truncation doesn’t happen till after 120k then its logical.