Document reading on LogicsDocBench benchmark leaderboard · SOTA2 Research

Document reading on LogicsDocBench

87.01Overall ScoreGemini-3-Pro

Evaluation Results

Method	Links
Gemini-3-Pro2026.03		87.01	8.15	84.86	84.33	90.66	14.68
MinerU2.52026.03		86.3	7.54	85.07	81.36	84.48	13.34
GLM-OCR2026.03		85.59	6.89	84.1	79.57	83.9	14.9
PaddleOCR-VL-1.52026.03		85.43	5.87	82.78	79.38	84.3	12.12
PaddleOCR-VL2026.03		85.09	5.81	83.8	77.27	81.6	14.04
Logics-Parsing-Omni2026.03		84.9	6.66	80.41	80.93	84.25	12.23
dots.ocr2026.03		84.51	6.29	81.6	78.21	82.49	11.53
Qwen3-VL-235B-A22B2026.03		84.47	8.17	84	77.58	81.92	14.05
Gemini-2.5-Pro2026.03		84.29	9.85	81.07	81.64	87.67	16.89
Qwen3.5-397B-A17B2026.03		81.09	8.68	80.37	71.57	78.48	14.5
Qwen3-VL-30B-A3B2026.03		78.94	13.4	81.52	68.69	75.33	17.18
MinerU2-pipeline2026.03		77.93	10.13	75.89	68.03	76.14	21.57
MonkeyOCR-pro-1.2B2026.03		77.57	16.53	79.85	69.39	74.77	27.88
GPT-5.22026.03		77.43	20.72	83.41	69.61	74.6	21.48
DeepSeek-OCR 22026.03		77.04	11.11	76.92	65.32	70.48	15.54
HunyuanOCR2026.03		76.11	15.04	71.62	71.75	76.47	20.36
InternVL35-241B2026.03		76.03	20.35	83.23	65.21	72	21.38
DeepSeek-OCR2026.03		74.45	16.5	76.9	62.94	66.76	21.89
Qwen3-Omni-30B-A3B2026.03		73.5	20.93	72.07	69.37	73.49	23.17
PP-StructureV32026.03		70.26	11.55	78.2	44.12	74.65	18.91