Tool and Model Comparison

10/10 = top within this comparison category · 8–9/10 = very strong · 6–7/10 = situational · 0/10 / – = not applicable

Tool / Model	OCR	Document Classification	Layout / Tables	Bounding Boxes	JSON / Structured Output	Local / Self-hosted	Speed / Efficiency	Adaptability to defined document classes	Production Readiness	Overall for
Gemini 2.5 Flash-Lite	8/10	10/10	8/10	8/10	10/10	0/10	10/10	10/10	10/10	8.5/10
Gemini 3.5 Flash-Lite	9/10	10/10	9/10	6/10	10/10	0/10	10/10	10/10	10/10	8.5/10
Qwen3-VL 4B	8/10	8/10	8/10	7/10	9/10	10/10	9/10	10/10	8/10	8.5/10
Qwen3-VL 8B	9/10	9/10	9/10	7/10	9/10	10/10	8/10	10/10	8/10	9/10
Qwen3-VL 30B-A3B	9/10	10/10*	9/10	8/10	9/10	10/10	8/10	10/10	8/10	9.5/10
Qwen3-VL 32B	9/10	10/10*	10/10	8/10	9/10	10/10	6/10	10/10	8/10	9.5/10
Qwen3-VL 235B-A22B	10/10	10/10*	10/10	9/10	9/10	9/10	3/10	10/10	8/10	8.5/10
Gemma 4 E2B	7/10	8/10	7/10	6/10	9/10	10/10	10/10	10/10	8/10	8/10
Gemma 4 E4B	8/10	9/10	8/10	7/10	9/10	10/10	9/10	10/10	8/10	8.5/10
Azure Document Intelligence Read/Layout	10/10	5/10	10/10	10/10	10/10	1/10	9/10	6/10	10/10	8.5/10
Azure Custom Classifier	5/10	10/10	8/10	6/10	10/10	1/10	9/10	9/10	10/10	8.5/10
PP-OCRv6 Medium	10/10	2/10	5/10	10/10	8/10	10/10	10/10	4/10	9/10	9/10 as OCR
PaddleOCR-VL	10/10	7/10	10/10	9/10	9/10	10/10	8/10	8/10	8/10	9/10
PP-StructureV3	9/10	5/10	10/10	9/10	10/10	10/10	8/10	7/10	9/10	8.5/10
Docling	7/10	5/10	10/10	8/10	10/10	10/10	7/10	7/10	9/10	8/10
MinerU	7/10	4/10	10/10	8/10	9/10	10/10	9/10	6/10	8/10	7.5/10
Xberg	7/10	7/10	9/10	9/10	10/10	10/10	5/10 in test	9/10	7/10	7.5/10
Tesseract	6/10	1/10	3/10	8/10	5/10	10/10	10/10	4/10	10/10	6.5/10
OpenCV	–	–	–	–	–	10/10	10/10	10/10	10/10	10/10 as Preprocessing
EmbeddingGemma	0/10	8/10 for text	0/10	0/10	9/10	10/10	10/10	10/10	8/10	8/10 as Pre-Classifier
PaddleX Custom Classifier	0/10	10/10 potential	5/10	0/10	9/10	10/10	10/10	10/10	9/10	9/10 with enough training data

Note: Qwen3-VL 30B/32B/235B classification scores are capability-based candidates and should still be validated against the golden dataset. 10/10 means top within this comparison, not universally perfect.