Comment by starkparker

I was more excited to hear about "structured Markdown" than the LLM OCR model, but the extent of it just seems to be tagging certain elements. It's useful in the LLM context but not as much outside of it.