3 ms·

Additionally, GP can use the word ‘concise’ (or similar) in their prompt if they want more concise output from a model.

• dymk 3mo ago
  That doesn’t mean “use fewer thinking tokens”. It might use more as it mulls over how to make its response concise.
  • tibbar 3mo ago
    effort level is a separate parameter, and is probably what they want