AI BriefingNumind Releases 4B VLM NuExtract3
Key point
Numind has released NuExtract3, a VLM specialized in document structuring and OCR based on Qwen3.5-4B.
Details
NuExtract3, released by Numind under the Apache-2.0 license, is a 4B-parameter VLM (Vision Language Model) based on Qwen3.5-4B.
It is optimized for extracting information from documents containing visual structure—such as complex PDFs, screenshots, forms, tables, and receipts—and converting it into Markdown or JSON format.
Key features include:
- Markdown conversion of document images and structured data extraction
- Precise data extraction using target JSON templates
- Ability to handle tables, forms, and pages with complex layouts
- Multimodal capability supporting both text and visual inputs
The model was trained for 3 days on an 8xH100 node and is the successor to the existing NuMarkdown-8B-Thinking.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Apple Releases LensVLM-9B for Processing Compressed Text Images