lightonai/LightOnOCR-2-1B
Primitive: /extract · Extract ·
LightOnOCR
📄 Paper | 📝 Blog | 🚀 Demo | 📊 Dataset | 📓 Finetuning
MultimodalMultilingual
Overview
Hardware: — drives latency, throughput & cost
| Size | 1.0B params |
|---|---|
| Tasks | /extract |
| License | apache-2.0 |
| Languages | en, fr, de, es, it, nl, pt, sv, da, zh, ja |
| Latency | 125.7 s |
| Throughput | 89 tok/s |
| Cost | $2.49 /1M tok |
Cost is approximate — computed from list GPU prices; your actual price depends on the provider you deploy SIE with.
Extraction
| Output kinds | Text |
|---|---|
| Inputs | image |
| Max sequence length | — |
Benchmarks
olmOCR-Bench
Document text extraction accuracy across arxiv math, old scans, multi-column layouts, and tables
Corpus: 1,403 Queries: 1,403
Quality
accuracy 0.7790
Performance L4 b1 c4
Compare (0)Compare →