OpenAI is now pricing wait times directly into compute costs: background tasks can choose 2-hour or 8-hour windows, with token rates 30% or 40% lower than the standard tier; for now, these figures come from only one source: OpenAI.
AI cost optimization has a new option—not switching to a cheaper model, but sending non-urgent tasks to a slow queue. This conclusion would no longer hold if OpenAI canceled the discounts or if the slow queue couldn’t reliably complete tasks within the promised window.
AI cost optimization has a new option—not switching to a cheaper model, but sending non-urgent tasks to a slow queue. This conclusion would no longer hold if OpenAI canceled the discounts or if the slow queue couldn’t reliably complete tasks within the promised window.
