[arnabdey0503]

Have you done benchmark againest other methods?

[mrkn1]

I haven't. But the evals of the underlying model are published here, including on Omnibench. https://huggingface.co/PaddlePaddle/PaddleOCR-VL-1.5