Adjust pricing for OpenAI flex tier

[nick-youngblut]

Feature request

I am getting the same token cost in my langsmith traces regardless of whether I use flex processing or the default service tier. It appears that langsmith is not accounting for flex pricing.

Motivation

Show the true token costs when using OpenAI flex pricing.

Also requested by Unify https://langchain.slack.com/archives/C07A5T75PPV/p1750969139514979

[baskaryan]

what are you using to call openai models? the sdk wrap_openai/wrapOpenAI wrappers or langchain ChatOpenAI?

[nick-youngblut]

what are you using to call openai models?

langchain ChatOpenAI: ChatOpenAI(model="o4-mini", service_tier="flex")