I am out of the loop with Cursor. Did they make their own LLM model or do you mean they just capture what is being sent to other models and that is valuable?
they probably just stole all private codebases cursor ever saw. That + the ARR more than justifies the valuation even without any internal model.
They have a few in models that powered the app, including one that applied edits. Recently they also started fine tuning Kimi models under the “composer” brand, and Composer 2.5 is a very cost effective coding model. But I suspect that the real value is in the distribution they have, which is what I primarily meant.