Agentic Document Extraction
For teams automating document extraction end to end.
Starter
/ month
Billed monthly
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / month
5,000
Overage
$0.10 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Most Popular
growth
/ month
Billed monthly
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / month
25,000
Overage
$0.09 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Enterprise
Billed monthly
Complete Unstract solution on the Cloud or Self-Hosted, for large teams or Enterprises. Enterprise-grade support and compliance.
Pages / month
Custom
Overage
Custom
Agentic Prompt Studio
✓
Compliance
SOC 2, GDPR, HIPAA, ISO 27001
Support
Priority, dedicated
Starter
/ month
Billed annually
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / year
60,000
Overage
$0.10 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Most Popular
growth
/ month
Billed annually
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / year
300,000
Overage
$0.09 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Enterprise
Billed annually
Complete Unstract solution on the Cloud or Self-Hosted, for large teams or Enterprises. Enterprise-grade support and compliance.
Pages / month
Custom
Overage
Custom
Agentic Prompt Studio
✓
Compliance
SOC 2, GDPR, HIPAA, ISO 27001
Support
Priority, dedicated
| Feature | Starter | Growth | Cloud Enterprise | On-Prem Enterprise |
|---|---|---|---|---|
| Core Platform | ||||
| Prompt Studio | ✓ | ✓ | ✓ | ✓ |
| Agentic extraction | ✓ | ✓ | ✓ | ✓ |
| Human-in-the-loop review | ✓ | ✓ | ✓ | ✓ |
| ETL pipelines + connectors | ✓ | ✓ | ✓ | ✓ |
| LLMWhisperer (included) | ✓ | ✓ | ✓ | ✓ |
| Volume & workflow | ||||
| Pages / month | 5,000 | 25,000 | Custom | Unlimited |
| Security & compliance | ||||
| Encryption at rest / in transit | ✓ | ✓ | ✓ | ✓ |
| SOC 2 Type II | x | x | ✓ | ✓ |
| ISO 27001 / HIPAA / GDPR + DPA | x | x | ✓ | ✓ |
| SSO / SAML / audit logs | x | x | ✓ | ✓ |
| Data residency in your VPC | x | x | x | ✓ |
| Support | ||||
| Slack community | ✓ | ✓ | ✓ | ✓ |
| Dedicated success manager | x | x | ✓ | ✓ |
| Uptime SLA | x | x | 99.9% | Self-managed |
The 14-day Unstract trial comes loaded with LLMWhisperer Free for document parsing, Azure OpenAI GPT-4o LLM with 1M tokens, an embedding model, and Postgres Vector DB. Sign up and start extracting on day one. No additional setup required.
After your trial, connect your preferred stack.
API Hub gives you ready-to-use extraction APIs for common documents like Bank Statements, Tables, and Invoices.
If you need the parsing API without the Unstract platform layer, check out the LLMWhisperer tab above.
Faq
Query not covered? Talk to us.
One page equals one physical page of a document. A 10-page PDF is 10 pages. A single image counts as one page. Multi-page scanned files (like TIFFs) count each frame as a page.
You can continue processing. Extra pages are billed at the overage rate listed on your tier. No surprise cut-offs. No throttling. If you go over month after month, a higher tier is often cheaper than paying overage. Talk to us.
No. If our system cannot process a page because of a technical error on our end (a timeout, a service issue, a file the parser cannot read), you will not be charged. You are charged for pages that process successfully, even if the output quality is not perfect. Extraction accuracy depends on the source document.
We never train on customer data. Cloud is encrypted in transit and at rest. Cloud Enterprise adds SOC 2, ISO 27001, HIPAA, and GDPR compliance plus a DPA. On-Premise Enterprise runs inside your own VPC, so your data never leaves your infrastructure. Full details in our Trust Center.
Yes. Prompt Studio projects, workflows, API deployments, and ETL pipelines are portable between deployments. Teams often start on Cloud to validate the fit, then move to On-Premise when volume, compliance, or data-residency requirements push them there. The reverse is possible if your needs change.
Layout Preserving Parser
Pay-as-you-go pricing. Start free at 100 pages a day. Pay for what you use after that. Pick a processing mode by document type.
Native Text
For digital PDFs
/ 1000 pages
$0.001 per page
Low Cost
For clean scans
/ 1000 pages
$0.005 per page
High Quality
For hard scans
/ 1000 pages
$0.01 per page
High Quality + Form/Table
For forms & tables
/ 1000 pages
$0.015 per page
| Capability | Native Text | Low Cost | High Quality | High Quality + Form/Table |
|---|---|---|---|---|
| Pricing | ||||
| Per 1,000 pages | $1 | $5 | $10 | $15 |
| Per page | $0.001 | $0.005 | $0.01 | $0.015 |
| Best input type | ||||
| Native text PDFs | ✓ | ✓ | ✓ | ✓ |
| High-quality scans | x | ✓ | ✓ | ✓ |
| Medium/low-quality scans | x | Basic | ✓ | ✓ |
| Handwritten documents | x | Basic | ✓ | ✓ |
| Smartphone-captured images | x | Basic | ✓ | ✓ |
| Forms with checkboxes / radio | x | x | x | ✓ |
| Capabilities | ||||
| Layout preservation | ✓ | ✓ | ✓ | ✓ |
| Line reproduction in output | x | ✓ | ✓ | ✓ |
| Line splitting strategy choice | ✓ | ✓ | ✓ | ✓ |
| Image preprocessing | x | ✓ | x | x |
| Rotation / skew compensation | x | x | ✓ | ✓ |
| Form-element detection (checkbox / radio ) | x | x | x | ✓ |
| Table detection | x | x | x | ✓ |
| Auto-repair PDFs | ✓ | ✓ | ✓ | ✓ |
| Language support | All Unicode | 120+ | 300+ | 300+ |
| Source Document Highlighting (add-on) | $0.01 / line coordinate lookup (across all modes) | |||
LLMWhisperer is included in your plan, no separate charge. The pricing above applies to standalone LLMWhisperer usage.
ADD-ON
$0.01/ line coordinate data lookup (across all modes)
Speed up manual reviews with source document highlighting. Generally implemented as a side-by-side view with the source document and the extracted data.
Free tier
No credit card required. Beyond 100 pages, pay for the pages you process at the per-page rate above.
LLMWhisperer On-Prem Enterprise
Data residency in your AWS, Azure, or GCP infrastructure. Enterprise-grade support and SLA for regulated use cases.
Faq
About LLMWhisperer. For Unstract platform questions, switch to the Unstract tab above. For anything else, talk to us.
It depends on your input documents. Choose Native Text for general PDFs with embedded text. It’s the fastest and cheapest. Low Cost for clean, high-quality scans of printed documents. High Quality for scans, handwriting and smartphone photos. High Quality Form + Table for documents with checkboxes, radio buttons, form fields, and tables.
There is no monthly cap. After the free 100 pages a day, pay-as-you-go kicks in at the per-page rate for your mode. Process as much or as little as you need. Rate limits apply at the API level (to keep the service fair for all users), but they sit above what production workloads need. If you are planning sustained high-throughput usage, talk to us and we will raise your limit.
No. If the API returns an error for a technical reason (timeout, unreadable file, service issue), that page is not charged. You are charged for pages that process successfully, even if the extracted text is not perfect. OCR accuracy varies with input quality and no API can guarantee flawless output on every document.
We never train on customer data. Documents sent to the API are encrypted in transit and at rest, processed, and discarded. If data residency or compliance rules out sending documents to our cloud, LLMWhisperer Enterprise lets you self-host inside your own AWS, Azure, or GCP VPC, so data never leaves your infrastructure. Full details in our Trust Center.
Agentic Document Extraction
For teams automating document extraction end to end.
Starter
/ month
Billed monthly
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / month
5,000
Overage
$0.10 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Most Popular
growth
/ month
Billed monthly
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / month
25,000
Overage
$0.09 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Enterprise
Billed monthly
Complete Unstract solution on the Cloud or Self-Hosted, for large teams or Enterprises. Enterprise-grade support and compliance.
Pages / month
Custom
Overage
Custom
Agentic Prompt Studio
✓
Compliance
SOC 2, GDPR, HIPAA, ISO 27001
Support
Priority, dedicated
Starter
/ month
Billed annually
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / year
60,000
Overage
$0.10 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Most Popular
growth
/ month
Billed annually
For teams of all sizes, running light document workflows. Single environment, limited page volume.
Pages / year
300,000
Overage
$0.09 / page
Agentic Prompt Studio
✓
Compliance
Standard
Support
Standard
Enterprise
Billed annually
Complete Unstract solution on the Cloud or Self-Hosted, for large teams or Enterprises. Enterprise-grade support and compliance.
Pages / month
Custom
Overage
Custom
Agentic Prompt Studio
✓
Compliance
SOC 2, GDPR, HIPAA, ISO 27001
Support
Priority, dedicated
| Feature | Starter | Growth | Cloud Enterprise | On-Prem Enterprise |
|---|---|---|---|---|
| Core Platform | ||||
| Prompt Studio | ✓ | ✓ | ✓ | ✓ |
| Agentic extraction | ✓ | ✓ | ✓ | ✓ |
| Human-in-the-loop review | ✓ | ✓ | ✓ | ✓ |
| ETL pipelines + connectors | ✓ | ✓ | ✓ | ✓ |
| LLMWhisperer (included) | ✓ | ✓ | ✓ | ✓ |
| Volume & workflow | ||||
| Pages / month | 5,000 | 25,000 | Custom | Unlimited |
| Security & compliance | ||||
| Encryption at rest / in transit | ✓ | ✓ | ✓ | ✓ |
| SOC 2 Type II | x | x | ✓ | ✓ |
| ISO 27001 / HIPAA / GDPR + DPA | x | x | ✓ | ✓ |
| SSO / SAML / audit logs | x | x | ✓ | ✓ |
| Data residency in your VPC | x | x | x | ✓ |
| Support | ||||
| Slack community | ✓ | ✓ | ✓ | ✓ |
| Dedicated success manager | x | x | ✓ | ✓ |
| Uptime SLA | x | x | 99.9% | Self-managed |
The 14-day Unstract trial comes loaded with LLMWhisperer Free for document parsing, Azure OpenAI GPT-4o LLM with 1M tokens, an embedding model, and Postgres Vector DB. Sign up and start extracting on day one. No additional setup required.
After your trial, connect your preferred stack.
API Hub gives you ready-to-use extraction APIs for common documents like Bank Statements, Tables, and Invoices.
If you need the parsing API without the Unstract platform layer, check out the LLMWhisperer tab above.
Faq
Query not covered? Talk to us.
One page equals one physical page of a document. A 10-page PDF is 10 pages. A single image counts as one page. Multi-page scanned files (like TIFFs) count each frame as a page.
You can continue processing. Extra pages are billed at the overage rate listed on your tier. No surprise cut-offs. No throttling. If you go over month after month, a higher tier is often cheaper than paying overage. Talk to us.
No. If our system cannot process a page because of a technical error on our end (a timeout, a service issue, a file the parser cannot read), you will not be charged. You are charged for pages that process successfully, even if the output quality is not perfect. Extraction accuracy depends on the source document.
We never train on customer data. Cloud is encrypted in transit and at rest. Cloud Enterprise adds SOC 2, ISO 27001, HIPAA, and GDPR compliance plus a DPA. On-Premise Enterprise runs inside your own VPC, so your data never leaves your infrastructure. Full details in our Trust Center.
Yes. Prompt Studio projects, workflows, API deployments, and ETL pipelines are portable between deployments. Teams often start on Cloud to validate the fit, then move to On-Premise when volume, compliance, or data-residency requirements push them there. The reverse is possible if your needs change.
Layout Preserving Parser
Pay-as-you-go pricing. Start free at 100 pages a day. Pay for what you use after that. Pick a processing mode by document type.
Native Text
For digital PDFs
/ 1000 pages
$0.001 per page
Low Cost
For clean scans
/ 1000 pages
$0.005 per page
High Quality
For hard scans
/ 1000 pages
$0.01 per page
High Quality + Form/Table
For forms & tables
/ 1000 pages
$0.015 per page
| Capability | Native Text | Low Cost | High Quality | High Quality + Form/Table |
|---|---|---|---|---|
| Pricing | ||||
| Per 1,000 pages | $1 | $5 | $10 | $15 |
| Per page | $0.001 | $0.005 | $0.01 | $0.015 |
| Best input type | ||||
| Native text PDFs | ✓ | ✓ | ✓ | ✓ |
| High-quality scans | x | ✓ | ✓ | ✓ |
| Medium/low-quality scans | x | Basic | ✓ | ✓ |
| Handwritten documents | x | Basic | ✓ | ✓ |
| Smartphone-captured images | x | Basic | ✓ | ✓ |
| Forms with checkboxes / radio | x | x | x | ✓ |
| Capabilities | ||||
| Layout preservation | ✓ | ✓ | ✓ | ✓ |
| Line reproduction in output | x | ✓ | ✓ | ✓ |
| Line splitting strategy choice | ✓ | ✓ | ✓ | ✓ |
| Image preprocessing | x | ✓ | x | x |
| Rotation / skew compensation | x | x | ✓ | ✓ |
| Form-element detection (checkbox / radio ) | x | x | x | ✓ |
| Table detection | x | x | x | ✓ |
| Auto-repair PDFs | ✓ | ✓ | ✓ | ✓ |
| Language support | All Unicode | 120+ | 300+ | 300+ |
| Source Document Highlighting (add-on) | $0.01 / line coordinate lookup (across all modes) | |||
LLMWhisperer is included in your plan, no separate charge. The pricing above applies to standalone LLMWhisperer usage.
ADD-ON
$0.01/ line coordinate data lookup (across all modes)
Speed up manual reviews with source document highlighting. Generally implemented as a side-by-side view with the source document and the extracted data.
Free tier
No credit card required. Beyond 100 pages, pay for the pages you process at the per-page rate above.
LLMWhisperer On-Prem Enterprise
Data residency in your AWS, Azure, or GCP infrastructure. Enterprise-grade support and SLA for regulated use cases.
Faq
About LLMWhisperer. For Unstract platform questions, switch to the Unstract tab above. For anything else, talk to us.
It depends on your input documents. Choose Native Text for general PDFs with embedded text. It’s the fastest and cheapest. Low Cost for clean, high-quality scans of printed documents. High Quality for scans, handwriting and smartphone photos. High Quality Form + Table for documents with checkboxes, radio buttons, form fields, and tables.
There is no monthly cap. After the free 100 pages a day, pay-as-you-go kicks in at the per-page rate for your mode. Process as much or as little as you need. Rate limits apply at the API level (to keep the service fair for all users), but they sit above what production workloads need. If you are planning sustained high-throughput usage, talk to us and we will raise your limit.
No. If the API returns an error for a technical reason (timeout, unreadable file, service issue), that page is not charged. You are charged for pages that process successfully, even if the extracted text is not perfect. OCR accuracy varies with input quality and no API can guarantee flawless output on every document.
We never train on customer data. Documents sent to the API are encrypted in transit and at rest, processed, and discarded. If data residency or compliance rules out sending documents to our cloud, LLMWhisperer Enterprise lets you self-host inside your own AWS, Azure, or GCP VPC, so data never leaves your infrastructure. Full details in our Trust Center.
Unstract adheres to the strict rules and regulations of various compliance authorities. Rest assured, we have policies, systems, and processes to ensure that your data is always safe, secure, and private.
Try it yourself
Get started with a 14-day Unstract trial, no credit card required. Or book a walkthrough with our engineers.
See Unstract in action with walkthroughs of core features and real extraction workflows.
Managed cloud, on-premise, or open-source. Unstract adapts to your infrastructure needs, so choose what works best for you.
Prompt engineering Interface for Document Extraction
Make LLM-extracted data accurate and reliable
Use MCP to integrate Unstract with your existing stack
Control and trust, backed by human verification
Make LLM-extracted data accurate and reliable
