Nanonets-OCR2-3B-AIO-GGUF

The Nanonets-OCR2-3B model is a state-of-the-art multimodal OCR and document understanding model based on the Qwen2.5-VL-3B architecture, fine-tuned for advanced image-to-markdown conversion with intelligent content recognition and semantic tagging. It can extract and transform complex document elements including text, tables (in markdown and HTML), LaTeX equations, flowcharts (as mermaid code), signatures, watermarks, checkboxes, and handwritten documents across multiple languages, supporting structured and context-rich outputs ideal for downstream AI processing. The model is 8-bit quantized for efficient inference, has about 3 billion parameters, a large 125K token context window, and supports visual question answering by providing direct answers from documents where applicable. Its design enhances document digitization workflows by unlocking structured data from unstructured documents, making it valuable in legal, medical, financial, and technical domains where accurate semantic extraction is crucial.

Model Files

Quants Usage

(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)

Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):