Batch processing - is it possible to process multiple user messages in parallel?

[yuamhd]

I was wondering if it's possible to process multiple messages in parallel?
I would like to be able to call guardrails like:

messages = ["Hello, what's up", "How to apply for gym?", "How to steal million dollars?"]
result = await self.guardrails.ainvoke({'input': messages})

I've registered Claude 3.5 Sonnet as custom LLM, and overriden the _transform_input_to_rails_format function. I know that the model supports batch processing, but when it gets to _agenerate function, it just calls model for last message. I've also tried everything else I could think of, but with little success.

Am I doing something wrong? Is there a way to achieve this?