Sold by: John Snow Labs
Deployed on AWS
Free Trial
Transform complex documents into structured, schema-compliant JSON using the top-ranked self-hosted OCR model. Built for enterprise document automation, JSL Vision OCR Structured LLM extracts data from PDFs, forms, tables, and scanned documents while keeping sensitive information within your AWS environment.
Overview
Vision OCR Structured LLM is an enterprise-grade vision-language model designed to convert complex documents into structured, application-ready data.
Unlike traditional OCR solutions that stop at text extraction, JSL Vision OCR Structured LLM understands document structure and produces schema-compliant JSON outputs that can be consumed directly by business applications, analytics platforms, RAG systems, and automation workflows.
Organizations can eliminate manual document processing, reduce custom parsing logic, and accelerate document-driven workflows by extracting structured information directly from PDFs, forms, reports, tables, and scanned documents.
The model is optimized for production environments where reliability matters. Through schema-aware decoding, it generates guaranteed-valid JSON outputs, eliminating malformed responses, post-processing pipelines, and costly validation workflows.
Key business benefits include:
- Reduce manual document review and data entry
- Automate extraction from forms, reports, invoices, and business documents
- Convert unstructured content into structured JSON for downstream systems
- Accelerate document ingestion, analytics, and workflow automation
- Improve consistency and reliability of extracted data
- Deploy securely within your AWS environment while maintaining full control of sensitive information
- Support compliance and governance requirements through self-hosted deployment
The model is particularly well suited for Intelligent Document Processing (IDP), document automation, financial workflows, healthcare documentation, regulatory reporting, claims processing, enterprise search, and AI-ready data preparation.
Designed for organizations that require both performance and operational control, the model delivers industry-leading structured document extraction while running efficiently on a single GPU, making advanced document intelligence accessible without complex infrastructure requirements.
Performance
- 0.714 JSON-Diff accuracy on OmniOCR - #1 OS, #5 overall in the JSL Vision Benchmark Series
- Superior performance on Schema constrained OCR: Claude Sonnet 4.5 (0.709), Holo2-30B-A3B (0.684), Qwen3-VL-8B (0.676), Pixtral-Large (0.670)
- 0.268 CER on FUNSD flat-text OCR (100 pages)
IMPORTANT USAGE INFORMATION:
After subscribing to this product and creating a SageMaker endpoint, billing occurs on an HOURLY BASIS for as long as the endpoint is running.
- Charges apply even if the endpoint is idle and not actively processing requests
- To stop charges, you MUST DELETE the endpoint in your SageMaker console
- Simply stopping requests will NOT stop billing.
This ensures you are only billed for the time you actively use the service.
Highlights
- Industry-Leading Performance: >>Runs efficiently on a single NVIDIA A10G (24 GB) GPU. >>32K context length for multi-page documents >>Supports PDF, PNG, JPG, and any image-convertible format
- Structured Extraction Excellence: >>Superior JSON generation from complex document layouts >>Excellent chart and data visualization comprehension >>Advanced table extraction with structure preservation >>Robust handling of nested tables and hierarchical data >>Reliable key-value extraction from challenging layouts >>Generates guaranteed-valid JSON outputs through schema-aware decoding. >>Eliminates post-processing, JSON repair, and schema-validation workflows.
- >>Converts PDFs, forms, reports, tables, and scanned documents into application-ready data. >>Keeps sensitive data within your AWS environment with no dependency on external AI APIs. >>Delivers deterministic, consistent outputs suitable for regulated and compliance-sensitive workflows. >>Optimized for healthcare, financial services, insurance, legal, government, and enterprise document processing.
Details
Sold by
Categories
Delivery method
Latest version
Deployed on AWS
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
Free trial
Try this product free for 15 days according to the free trial terms set by the vendor.
Pricing is based on actual usage, with charges varying according to how much you consume. Subscriptions have no end date and may be canceled any time.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimension | Description | Cost/host/hour |
|---|---|---|
ml.g5.2xlarge Inference (Batch) Recommended | Model inference on the ml.g5.2xlarge instance type, batch mode | $9.98 |
ml.g5.2xlarge Inference (Real-Time) Recommended | Model inference on the ml.g5.2xlarge instance type, real-time mode | $9.98 |
ml.g5.12xlarge Inference (Batch) | Model inference on the ml.g5.12xlarge instance type, batch mode | $9.98 |
ml.g5.12xlarge Inference (Real-Time) | Model inference on the ml.g5.12xlarge instance type, real-time mode | $9.98 |
ml.g5.xlarge Inference (Batch) | Model inference on the ml.g5.xlarge instance type, batch mode | $9.98 |
ml.g5.4xlarge Inference (Batch) | Model inference on the ml.g5.4xlarge instance type, batch mode | $9.98 |
ml.g5.8xlarge Inference (Batch) | Model inference on the ml.g5.8xlarge instance type, batch mode | $9.98 |
ml.g5.xlarge Inference (Real-Time) | Model inference on the ml.g5.xlarge instance type, real-time mode | $9.98 |
ml.g5.4xlarge Inference (Real-Time) | Model inference on the ml.g5.4xlarge instance type, real-time mode | $9.98 |
ml.g5.8xlarge Inference (Real-Time) | Model inference on the ml.g5.8xlarge instance type, real-time mode | $9.98 |
Dimensions summary
You pay by the hour based on the AWS instance you run the software on. Pricing spans five g5 instance sizes, from ml.g5.xlarge up to ml.g5.12xlarge. Larger instances carry more GPU and compute, so the hourly rate rises with size. Each size also splits into two modes: batch, which processes documents in scheduled jobs, and real-time, which handles requests as they arrive. The ml.g5.12xlarge offers both modes, while the other sizes list batch and real-time separately. You are billed only for the hours you use, with no upfront commitment.
Top-of-mind questions for buyers
Batch mode processes documents in scheduled jobs, so you spin up an instance, run a set of documents, then stop. Real-time mode keeps an endpoint running to handle requests as they arrive. Batch suits periodic bulk processing; real-time suits live, on-demand extraction from applications.
You pay per host-hour only while the instance runs, with no upfront commitment. Stopped instances stop accruing software charges, though underlying AWS storage may still bill separately. Usage is metered by the hour, and you can run parallel instances, each metered independently.
No document, page, or character limits apply. You are billed only for instance host-hours, not for volume processed. Throughput depends on the instance size you pick and your document complexity, so a larger g5 instance processes more within the same hour.
www.johnsnowlabs.com
Helpful?
Vendor refund policy
No refunds are possible.
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
Amazon SageMaker model
An Amazon SageMaker model package is a pre-trained machine learning model ready to use without additional training. Use the model package to create a model on Amazon SageMaker for real-time inference or batch processing. Amazon SageMaker is a fully managed platform for building, training, and deploying machine learning models at scale.
Deploy the model on Amazon SageMaker AI using the following options:
Version release notes
Model Optimization
Additional details
Inputs
- Summary
1. Chat Completion
Example Payload {
"model": "/opt/ml/model",
"messages": [
{"role": "system", "content": "You are a helpful medical assistant."},
{"role": "user", "content": "What should I do if I have a fever and body aches?"}
],
"max_tokens": 1024,
"temperature": 0.6
}For additional parameters:
ChatCompletionRequest OpenAI Chat API
2. Text Completion
Single Prompt Example {
"model": "/opt/ml/model",
"prompt": "How can I maintain good kidney health?",
"max_tokens": 512,
"temperature": 0.6
}
Multiple Prompts Example {
"model": "/opt/ml/model",
"prompt": [
"How can I maintain good kidney health?",
"What are the best practices for kidney care?"
],
"max_tokens": 512,
"temperature": 0.6
}
Reference:3. Image + Text Inference
The model supports both online (direct URL) and offline (base64-encoded) image inputs.
Online Image Example { "model": "/opt/ml/model", "messages": [ {"role": "system", "content": "You are a helpful medical assistant."}, { "role": "user", "content": [ {"type": "text", "text": "What does this medical image show?"}, {"type": "image_url", "image_url": {"url": "https://example.com/image.jpg "}} ] } ], "max_tokens": 2048, "temperature": 0.1 } Offline Image Example (Base64) { "model": "/opt/ml/model", "messages": [ {"role": "system", "content": "You are a helpful medical assistant."}, { "role": "user", "content": [ {"type": "text", "text": "What does this medical image show?"}, {"type": "image_url", "image_url": {"url": "data:image/jpeg;base64,..."}} ] } ], "max_tokens": 2048, "temperature": 0.1 } Reference:
4. Structured Output (JSON Schema)
Force the model to output valid JSON matching a specific schema using response_format.
Example with Schema
{ "model": "/opt/ml/model", "messages": [ { "role": "system", "content": "Extract patient information as JSON." }, { "role": "user", "content": "Patient John Doe, age 45, has hypertension." } ], "temperature": 0.0, "max_tokens": 512, "response_format": { "type": "json_schema", "json_schema": { "name": "patient_info", "strict": true, "schema": { "type": "object", "required": ["name", "age", "conditions"], "properties": { "name": {"type": "string"}, "age": {"type": "integer"}, "conditions": { "type": "array", "items": {"type": "string"} } } } } } }
Reference:
Important Notes:
- Streaming Responses: Add "stream": true to your request payload to enable streaming
- Model Path Requirement: Always set "model": "/opt/ml/model" (SageMaker's fixed model location)
- Input MIME type
- application/json
https://github.com/JohnSnowLabs/spark-nlp-workshop/tree/master/products/sagemaker/models/jsl_vision_ocr_structured_light/inputs
https://github.com/JohnSnowLabs/spark-nlp-workshop/tree/master/products/sagemaker/models/jsl_vision_ocr_structured_light/inputs
Support
Vendor support
For any assistance, please reach out to support@johnsnowlabs.com .
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
IBM Granite Vision 3.3 2B is a compact multimodal model for accurate document image understanding, segmentation, and visual question answering.
Compact, 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
Searce’s LLM powered Intelligent Document Processing solution combined with AWS Professional Services is an automation bridge that transforms raw and unstructured documents into structured form documents using advanced OCR, NLP and Vision-based LLMs.
Nanonets is an intelligent document processing API that converts PDFs and images into structured markdown, JSON, or extracted fields. Purpose-built for LLM and RAG applications with industry-leading table extraction, multi-language OCR, and confidence scoring. SOC 2, GDPR, and HIPAA compliant.
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.