Google is introducing two new inference tiers to the Gemini API, Flex and Priority, to balance cost and latency.
