Vision OCR LLM
John Snow Labs Inc · Software Development
Certification per Microsoft Marketplace.
Evidence tier Source Confirmed · 9 captures on record
What the publisher says
As described on Microsoft Marketplace.
The Vision OCR LLM is an enterprise-grade OCR-specialized vision-language model engineered for state-of-the-art grounded OCR in production document workflows. It is the right model when text recognition AND text location both matter: medical de-identification, form-field extraction, compliance redaction, document anonymization, and any pipeline that needs to act on a specific word at a specific position on a specific page
The model emits text along with precise word-level bounding-box coordinates in a single inference pass, with no two-stage detection-then-recognition pipeline to maintain, achieving state-of-the-art results across every major OCR benchmarks.
Show the rest of the publisher’s description (20 more lines)
Unlike traditional OCR solutions that only return text, the model is optimized for reading text and returning precise word-level bounding boxes in a single inference pass.
Key capabilities and Ideal Use Cases
- OCR and document understanding for PDFs, images, forms, and scanned documents
- Medical de-identification (PHI redaction with precise coordinates)
- Form-field extraction (mapping values to specific page regions)
- Compliance auditing (which text was flagged, where on the page)
- Document anonymization (region-level masking and blurring)
- Multilingual document processing, table and formula recognition, handwritten text
In independent benchmark evaluations covering leading OCR and vision-language models, John Snow Labs Vision OCR LLM achieved the highest ranking among self-hosted models and outperformed multiple well-known open-source and commercial alternatives on structured document extraction tasks. The model is specifically designed for organizations that require accurate document intelligence while maintaining security, compliance, and operational control
Performance
860 on OCRBench (state-of-the-art for models under 3B parameters)
94.10 overall on OmniDocBench with 0.042 text edit distance, 94.73 formula, 91.81 table
85.21 on Wild-OmniDocBench (degraded scans with folds and lighting changes)
91.03 on DocML multilingual document parsing across 14 non-English non-Chinese languages
92.29 cards, 92.53 receipts, 92.87 video subtitles on information extraction
0.9574 Table TEDS, 0.9706 Formula CDM (English)
0.039 BBox CER on FUNSD - #1 of 15 models in the JSL Vision Benchmark Series
4.7x lower CER than Tesseract 5.5, 6.1x lower than EasyOCR on the same FUNSD benchmark
100% parse rate - valid bounding-box output produced for every page
Built for organizations that require security, control, and high-quality structured outputs, the Vision OCR LLM enables enterprises to unlock value from document repositories while reducing operational costs and accelerating automation initiatives.
Agent build and provenance
See the full provenance
The layer-by-layer build, the evidence behind each claim, the risk basis and the cross-marketplace links are open to any account. Some rows are disclosed, some the source leaves Unknown; a free account shows you which.
Compliance
- FedRAMPConfirmedNot listed90%, registry-checkedNo FedRAMP Marketplace entry matched this vendor's domain, checked 2026-08-27registry recordas observed 2026-08-27
Confirmed means matched to a public authoritative registry. Claimed means the vendor or its listing states it, not yet cross-checked. A framework not shown was not found in any source we hold, which is not evidence against it. Not listed means a scoped registry check found no match for this vendor's domain: a No is a scoped registry check, not a compliance judgment. Confidence bands: 95% domain-verified, 90% registry-checked, 80% self-attested, 70% weak signal. Self-attested items marked “vendor's site” are gathered from the vendor's own website and are not verified by us.
Sources
Publisher resources
4 linksLinked repositories
Unknown means this listing does not publish a repository. It is not a statement that the code is closed, and a linked repository is not a claim that the publisher wrote it: the registry computes that relationship privately and does not publish it.
Evidence risk is the share of the build you cannot see before you deploy, not a security rating. Sign in to see the layer-by-layer basis for this band.