[hauntsaninja]

Reasoning effort is denominated in tokens, not time, so no difference beyond slowness at heavy load

(I work at OpenAI)