Before the recent updates it was easy to see if we were using a reasoning model as the selected label showed a brain icon next to the model (eg sonnet 4.5 thinking) but now we do not know if we are using sonnet 4.5 (1x request) or sonnet 4.5 thinking (2x the requests).
I think a visual element should be there to differentiate if we are using a 1x model Vs 2x Vs MAX mode.
This is sonnet “thinking” in max mode
There was a bug report submitted for it that has been closed, but the issue still persists.
@david_frechette It also might be helpful to edit your post and make this a “Bug Report” instead of feedback.
As a side note: with the new pricing policy, I am not sure if sonnet 4.5 thinking is more expensive than non thinking (if so, it may be negligible). I have ran identical requests through both and they appeared to do similar work and the cost per tokens was the same. They should still be showing the “brain” icon regardless.
It would be nice if Cursor had an MCP server that could provide usage and cost data broken down by agent conversation/task.