I pay for Cursor but the comments here made me question the incentives. They have to cut spending. The byo api keys don’t suffer from that. Maybe we need evals

I pay for Cursor but the comments here made me question the incentives. They have to cut spending. The byo api keys don’t suffer from that. Maybe we need evals for these tools? There are benchmarks for coding abilities (swe bench?), would be cool to set these agentic IDEs.