What breaks in document AI

Large and long documents, lost pages

14 of 14 quotes

• “I want to summarize a large pdf files approximately 200 pages using Gemma 4 E4B on LM Studio app but it cannot read the file it start reading and remained at 0% not moving”

• “Marker failed to export the embedded image on 5 of 36 pages”

• “When processing a large Thai PDF containing 104 pages as a single MinerU job, OCR processing appears to complete successfully for all pages, but the task fails during the final stage”

• “My first attempt was using LlamaParse. It completely failed for this specific document. It merged pages together into 624 massive chunks, missed the Article boundaries, and ingested all the footnotes.”

• “I have an 80Mb PDF, which has 184 pages of scanned grayscale images.”

• Hit this one too? We are building for it.

• “The pipeline continues but layout analysis is silently skipped for failed pages.”

• “When I use Deepdoc to parse long page numbers (over 1000 pages), the error message is as follows”

• “We used Reducto and it did struggle with long documents. As we process financial documents going over 300+ pages using Gemini 3 Flash is producing high accuracy extracts super fast.”

• “I’ve found it challenging to get clean copies of long documents (books).”

• “Processing that same page on my self-hosted setup produces: - empty output JSONL - no markdown output - pipeline log says No document text”

• “My pdf is 492 pages long, and I find that the OCR process got stuck on this state for about 10 mins”

• “When parsing certain PDF documents, some pages fail to parse and are silently skipped.”

• “During embedding, some chunk exceeds 8192 and causes an error. However, if I parse the same PDF with Deepdoc, it succeeds.”

• “Raptor is failing for this file with error INVALID_REQUEST”

• Hit this one too? We are building for it.