The Sonnet 3.7 'Agent' implementation in Cursor is terrible. It takes 10x longer to produce a worse response. I don't understand if this is a cost saving measure to reduce token count, but it is really bad.
Thankfully, switching to Sonnet 3.7 in 'Ask' mode has been essentially fine and is comparable to how Sonnet 3.5 performed for me.
Thankfully, switching to Sonnet 3.7 in 'Ask' mode has been essentially fine and is comparable to how Sonnet 3.5 performed for me.