VikParuchuri /surya
OCR, layout analysis, reading order, table recognition in 90+ languages
★ 21,592⑂ 1,572👁 120⑃ 261● 309▤ 89
下载:Surya v0.24.0 — Zero-PyTorch Lightweight Architecture, Interactive Web Viewer (surya_gui) & Agent-Ready CLISurya-0.24.0 ↓0.24.02026-11-13
+2
日增长
+15
周增长
+91
月增长
21,592
stars
1,572
forks
120
watchers
261
prs
309
issues
89
releases
README
Datalab State of the Art models for Document Intelligence Surya Surya is a 689M param document OCR and layout VLM with these features: - Accuracy - scores 85.9 on OmniParseBench, the highest of the open models under 1B parameters we have scored, and 83.7% on olmOCR-bench - Speed - throughput of 4.5 pages/s on one H100 at 192 DPI, 1.7x Surya 2 with MTP (see Throughput) - Multilingual - scores 74.9% on an internal benchmark set of 91 languages (more here) - Layout analysis (Table, Picture, SectionHeader, Equation, Form, etc.) with reading order and token-count estimates - Tables as HTML (, , , ) with merged cells as colspan / rowspan - Zero heavy local ML dependencies - surya-ocr is a lightweight client/orchestrator that talks to vllm (Docker / Podman), llama-server (llama.cpp), or mlx_vlm.server (Apple Silicon) Try Datalab's Managed Platform Our managed platform uses Surya, our highest accuracy model Chandra, and other models. The Datalab API turns documents into verifiable knowledge your agents act on. Every result traces back to its page and position, and one API can parse, extract, fill and edit documents. Get started with about 5,000 free convert pages every month (work email) — sign up (takes under 30 seconds) or try our free public playground. Model Information | Layout | OCR | |:------------------------------------------------------------------:|:-----------------------------------------------------------------------:| | | | Surya is named for the Hindu sun god, who has universal vision. Examples Each row links to annotated views of the same page: OCR, layout, and reading order (Surya 2.5). | Name | Original | OCR | Layout | Order | |-------------------|:-----------------------------------:|------------------------------------------:|---------------------------------------------:|------------------------------------------------:| | Newspaper | Image | Image | Image | Image | | Reference book | Image | Image | Image | Image | | Math paper | Image | Image | Image | Image | | Government form | Image | Image | Image | Image | | Handwritten letter | Image | Image | Image | Image | | Corporate doc | Image | Image | Image | Image | | Japanese tax form | Image | Image | Image | Image | The example pages are public pages from OmniParseBench. The newspaper (The Alpha, 1881), reference book and math paper are from olmOCR-bench (Ai2, ODC-BY 1.0). The government form is a US federal form (public domain). The handwritten letter is from the Library of Congress, Manuscript Division, Carrie Chapman Catt Papers (public domain). The corporate page is from ParseBench (Apache 2.0). The Japanese form is made from a form of the Nat
Top issues
#574[FEAT] Apple throughput0 comments · 2026-11-13#573Verify evals on Papers with Code0 comments · 2026-11-13
Latest release
Surya v0.24.0 — Zero-PyTorch Lightweight Architecture, Interactive Web Viewer (surya_gui) & Agent-Ready CLI v0.24.0 2026-11-13Surya v0.24.0 — Zero-PyTorch Lightweight Architecture, Interactive Web Viewer (surya_gui) & Agent-Ready CLI Surya 0.24.0 completes the transition to Surya 2.5 (datalab-to/surya-ocr-2.5) — our 689M-parameter vision-language model for full-page & per-block OCR, layout analysis & reading order, inline & block LaTeX math (), and HTML table recognition ( with colspan / rowspan) across 90+ languages: - 85.9 on OmniParseBench (highest of any open model under 1B parameters) and 83.7% on olmOCR-bench. - 4.5 pages/s on a single H100 at 192 DPI (6.2 pages/s at 96 DPI) — 1.7× faster than Surya 2 and 3.2