1. Marker

Marker is a OCR + LLM lib with propietary table transformer

some key features:

• llm mode that augments marker with models like gemini flash
• improved math, w/inline math
• links and references
• better tables and forms

LLM mode iterates on marker output for certain blocks. You can use gemini, or local models via ollama. More models coming soon.

Marker + llms is faster and hallucination-free vs using llms alone. Here marker

• gemini flash beats gemini flash alone on a fintabnet benchmark.

![marker

• gemini](https://pbs.twimg.com/media/GkKzzzbWYAERp56?format=jpg&name=small)

Table improvements are:

• new table recognition model
• table merging across pages
• math and formatting inside tables
• output in html, markdown, or json

Up next is:

• More formats (docx, pptx, xlsx, etc)
• Improved layout detection on scientific papers, engineering documents, newspapers
• Structured data extraction

2. llamaindex / llamaparse

It is really good at the following:

✅ Broad file type support: Parsing a variety of unstructured file types (.pdf, .pptx, .docx, .xlsx, .html) with text, tables, visual elements, weird layouts, and more. ✅ Table recognition: Parsing embedded tables accurately into text and semi-structured representations. ✅ Multimodal parsing and chunking: Extracting visual elements (images/diagrams) into structured formats and return image chunks using the latest multimodal models. ✅ Custom parsing: Input custom prompt instructions to customize the output the way you want it.

Overall more easily integrated into LLAMAIndex if needed.