Running two LLMs in parallel in streamlit

I am testing a few LLM setups for my task which I want to compare in A/B tests.
I would like to show outputs generated by two LLMs to compare them side-by-side. As this generation takes some time, I would not like to wait for one to finish, and then start the second one.
Ideally - I would like to have two column with steaming tokens (as they arrive from LLM) in the same time.

I did not find clear guidelines about parallelism in streamlit.
Should I use asyncio, threading?
Or maybe there is some streamlit-specific way to handle two LLM streams?

Please advise