Multilingual Document Parsing on MORE benchmark leaderboard · SOTA2 Research

Multilingual Document Parsing on MORE

91.9Overall AccuracyHunyuanOCR-1.5

Evaluation Results

Method	Links
HunyuanOCR-1.52026.07		91.9	91.31	91.1	80.77	99.1	92.66	96.48
Gemini 32026.07		91.61	95.39	90.27	81.02	93.05	94.31	95.63
PaddleOCR-VL-1.62026.07		89.88	90.28	89.16	76.46	97.47	92.86	93.05
PaddleOCR-VL2026.07		87.96	90.99	91.11	61.11	96.29	93.04	95.19
GLM-OCR2026.07		85.75	87.31	89.29	82.48	95.83	67.12	92.48
Unlimited-OCR2026.07		84.9	86.75	92.22	50.89	97.45	85.95	96.17
dots.ocr2026.07		84.31	94.45	90.77	39.81	95.38	88.26	97.18
Qwen2.5-VL2026.07		83.93	89.36	84.48	68.27	86.69	92.54	82.23
Qwen3-VL2026.07		83.56	92.02	65.45	65.21	92.38	93.76	92.53
DeepSeek-OCR2026.07		82.91	85.27	75.67	61.63	92.26	88.26	94.36
MinerU 2.52026.07		48.85	27.12	73.29	33.83	72.41	21.61	64.81