We're sharing a new method for scoring models on agentic coding tasks. Here's how models in Cursor compare on intelligence and efficiency:

We use a combination of offline benchmarks and online evals to measure model quality. This makes results more useful, especially as public benchmarks are increasingly saturated.

Learn more: cursor.com/blog/cursorben…

@cursor_ai please share the repo where the code for the tests resides, I want to reproduce it

@cursor_ai i’m confused, why does it say lower score is better in the final figure for online evals?

@cursor_ai Hope you get bankrupt at some point with all your practice screwing your userbase with lies. Sad that it has to come that far! At least be honest in your communication lol. OH VEY