Ran out sonnet 4.0 in 35 requests - any ideas on how to to get back old pricing option?

hi @umerqr

Cursor does not track requests but token and model usage which depends on details reported by AI providers through API.

Your screenshot shows high token consumption.

This may mean

  • Very heavy coding usage
  • Large context used

Recommendations

  • Use thinking models only when there is a clear requirement to use them as they do not have much higher quality output than non-thinking models, but consume much more tokens.
  • Check and avoid attaching too much context like files, rules, MCPs,…
  • Keep chats focused on single tasks and short.

For more see following, incl. how to optimize your usage: