How to simply route requests with the same request id to the same model instance?

[fighterhit]

How can I simply route all inference requests with the same request id to the same model instance, and then execute inference using dynamic_batching? It sounds like this can be achieved by using a stateful model to change the request id to a sequence id, but it feels too complicated and requires additional control input, because I only need the ability to route to the same instance and dynamic batches. Is there an easy solution? Thanks!