view article Article LightOnOCR-1B: The Case for End-to-End and Efficient Domain-Specific Vision-Language Models for OCR lightonai โข Oct 23, 2025 โข 74