GitLab Transcend - Intelligent orchestration
Mini toolsLightOnOCR 2
By LightOn
LightOnOCR-2-1B is an OCR-focused vision-language model tuned for accurate document understanding rather than simple text extraction. LightOn says the second generation improves French, arXiv, and scan coverage, LaTeX handling, and normalization quality, while achieving state-of-the-art performance on OlmOCR-Bench. The model card also highlights speed and efficiency, including 5.71 pages per second on a single H100, support for 11 languages, and variants for bounding-box prediction and fine-tuning.
Text
Gen 7
Overview
LightOnOCR-2-1B is LightOn’s flagship 1B-parameter OCR vision-language model for turning PDFs, scans, and images into clean, naturally ordered text. It is designed as an end-to-end OCR system that handles tables, forms, receipts, multi-column layouts, and math without relying on brittle external pipelines.
About LightOn
We are pioneers in Generative AI, building secure and customizable solutions for enterprises to improve productivity and competitiveness.
Company Size:
52
Website:
lighton.ai
Tools using LightOnOCR 2
No tools found for this model yet.
Last updated: April 6, 2026
