Is there a way to have fast lower quality processing? <5s

[simjak]

I have a case where the documents don't need OCR, just simple text documents. However, the latency is important.
e.g. if I use Jina AI reader:
It parses this document within a few seconds https://arxiv.org/pdf/2305.10250 to markdown

https://r.jina.ai/https://arxiv.org/pdf/2305.10250