support flex processing for OpenAI

[rubenvereecken]

Description

Flex processing allows 50% cheaper requests at the cost of latency. Only available for o3 and o4-mini models.

It would require a new option providerOptions.openai.service_tier that can be set to flex.

AI SDK Version

No response

[lgrammel]

cc @dancer

[dancer]

Thank you for waiting, it will be added to v5 @rubenvereecken

[rubenvereecken]

Awesome, nice one @lgrammel !