Granite 4.0 3B Vision: Compact Multimodal Intelligence for Enterprise Documents

Summary

Granite 4.0 3B Vision is a compact vision-language model designed to extract information from enterprise documents, including tables, charts, forms, and semantically meaningful key-value pairs. It combines a million-scale ChartNet dataset with a code-guided synthesis pipeline, DeepStack Injection for routing semantic and high-resolution visual features to different layers, and a LoRA adapter architecture built on Granite 4.0 Micro. The modular design supports standalone image processing, text-only fallback, and integration with Docling for multi-page document pipelines. Reported evaluations show 86.4% on Chart2Summary, leading the evaluated models, second place on Chart2CSV at 62.1%, leading scores on the listed table benchmarks, and 85.5% zero-shot exact-match accuracy on VAREX.

Takeaways

1. 1

   ChartNet contains 1.7 million chart samples spanning 24 chart types and six plotting libraries, with synthetic, human-annotated, and real-world subsets.

2. 2

   DeepStack Injection separates semantic and spatial visual features across model layers, supporting layout-sensitive tasks such as table extraction, chart interpretation, and key-value parsing.

3. 3

   The model achieved 92.1 on cropped PubTables-v2, 79.3 on full-page PubTables-v2, 64.0 on OmniDocBench, 88.1 on TableVQA, and 85.5% zero-shot exact match on VAREX.