LightOnOCR-3 Reads 16th-Century Manuscripts and Ancient Greek Without Training
IgorCarron · x · 2026-11-02
French company LightOn released LightOnOCR-3, an end-to-end document understanding model (predecessor: 4.5M+ downloads). The poster's use case: reading and "digitizing" 16th-century Florentine manuscripts, Ancient Greek and Arabic without task-specific training — e.g., mining 3,000 un-digitized heterogeneous manuscript pages.
Technically: #1 on FRBench-pdf2md, #2 on OlmOCR-Bench, #1 among open-weight models on ParseBench; it also reconstructs document layout (paragraphs/tables with bounding boxes), describes images, and extracts structured data from charts. Open source and compact.
Related event: LightOn Releases LightOnOCR-3 for End-to-End Document Understanding(2 posts)→
More from Models
- ICML paper: finite context windows impose impossibility results for adversarial regret — burny_tech · 2026-11-02
- SenseTime open-sources SenseNova-Loop-9B: weight-shared 56-layer loop, 256K context, Apache 2.0 — AdinaYakup · 2026-11-02
- User says Gemini excels at long-form fanfic, producing stories up to 29 minutes of reading — Ok-Jello3739 · 2026-11-02
- Sam Altman calls tokens 'a kind of fake unit' that doesn't measure business value — rohanpaul_ai · 2026-11-02
- 'You Can Feel the Gradient Slopes': Dev Contrasts Benchmaxxed RL Models vs General Ones — theteknosaur · 2026-11-02
- LightOnOCR-3 Launches: Tops FRBench, Adds Layout Structure and Chart Extraction — IgorCarron · 2026-11-02