How to speed up the Assistant?

I´m using a custom Assistant which should answer product-related questions.
Due to token limitation I had to split the product feed into 20 different JSON files.
The Assistant uses retrieval function to find the best answers from the files.
The answers are relatively good, but it takes 10+ seconds to generate an answer.
I´m in Tier 3, is there a way to improve the speed of response?

10 seconds is not too bad, I have had Assistant impl that take 1-2 minutes to answer.

10 seconds is normal given that there is no streaming, and you need to wait for retrieval and message population.

Now, generally, to make it faster, use 3.5 rather than 4, use easier to read file formats, and give short instructions that don’t force the assistant to perform unnecessary tasks. Ask it to be concise so as to avoid some unneeded response words.