Table Parsing on OCRBench V2 benchmark leaderboard · SOTA2 Research

Table Parsing on OCRBench V2

91.39TEDSStrucTab

Evaluation Results

Method	Links
StrucTab2026.06		91.39	94.4
TRivia-3B2026.06		90.76	94.03
StrucTab-SFT2026.06		89.33	91.69
Gemini 2.5 Pro2026.06		88.94	89.47
MinerU 2.52026.06		87.13	90.62
Qwen3-VL-A22B-Ins.2026.06		84.12	88.15
DeepSeek-OCR2026.06		82.64	87.33
Kimi K2.5 (no-think)2026.06		82.51	89.71
HunyuanOCR2026.06		82.34	87.74
dots.ocr2026.06		82.04	86.27
Qwen2.5-VL-72B-Ins.2026.06		81.33	86.58
GPT-52026.06		79.91	88.69
InternVL3.5-A28B2026.06		79.5	85.81
PaddleOCR-VL2026.06		79.29	83.93
FD-RL2026.06		76.95	82.39
MonkeyOCR-pro-3B2026.06		76.81	79.69
POINTS-Reader2026.06		73.96	82.13
GPT-4o2026.06		70.51	79.55
Seed-1.8 (no-think)2026.06		69.91	77.95
UniTable2026.06		67.73	78.65
MonkeyOCR-pro-1.2B2026.06		67.21	72.81