Upload a Persian document image (printed or handwritten) and the Bina 0.1 Koochik model — a 665M-parameter vision-language model fine-tuned from Surya OCR 2 for Persian — transcribes it to text.