Best practice to setup rate limit within a LangGraph Cloud app

[arthberman]

We have multiple assistants deployed on LangGraph Cloud, we use the LangGraph API to interact with. We use sub graph and ReAct nodes.
What is the best practice to handle our API rate limit (OpenAI, Anthropic and Bedrock), as we are using the same API key for multiple assistants so we share the same tier ?
How to share a InMemoryRateLimiter through multiple assistant ? Is it the good approach ?

[ESJavadex]

@arthberman Did you find any interesting solution?