Nanonets bills per block run, not per page: $0.02 for a simple operation, $0.10 for standard AI, $0.30 for complex AI. Starter is $50 in credits, then $100 a month for 100 credits. Growth and Enterprise are quote-only. By Nanonets' own example, a typical invoice runs 4 to 6 blocks and lands under $2 end to end.
That unit does not convert to the per-1,000-pages rate every other vendor quotes. Here is the honest bridge. Last updated July 2026.
Upload a document to extract
Drop files here or click to upload
Up to 50 files
Uploading...
Before you count blocks in someone's workflow builder, drop in the document you meant to test and see what finished, validated fields look like.
Nanonets charges $0.02 for a simple block run, $0.10 for a standard AI run, and $0.30 for a complex AI run. A block is one step in a workflow: extract the data, classify the document, format a field, post to an integration. Run five steps over a document and you have five billable runs. Nanonets' own pricing page says a typical invoice workflow runs 4 to 6 blocks and comes to under $2 per invoice end to end, which is the most useful sentence on it. Starter costs $100 a month for 100 credits after $50 of starting credit; Growth and Enterprise publish nothing and route you to sales. Now the part that trips up every spreadsheet: nobody else in this market prices this way. Azure, AWS, Google, and Mistral all quote a rate per 1,000 pages. Nanonets quotes a rate per step. Those units do not convert, and comparing them without saying so produces a number that is off by a factor of fifty and means nothing.
Take Nanonets at its own word. Four to six complex blocks at $0.30 each puts a single-page invoice at $1.20 to $1.80. Scale that to 1,000 invoices and you are looking at roughly $1,200 to $1,800. Azure custom extraction charges about $30 per 1,000 pages. On the face of it Nanonets costs forty to sixty times more, and that comparison is close to worthless.
Azure's $30 buys one model call that hands back fields. Nanonets' $1.20 buys the document classified, the fields extracted, the values formatted and validated, and the result posted into your ERP. Azure charges nothing at all for the four steps it does not perform. You will pay for those anyway, in engineering salary, and you will keep paying every time a format changes.
A real workflow is also rarely all complex blocks. Two complex runs plus three standard runs is $0.90 a document, not $1.50. Count your actual steps and tiers before you accept the headline.
Rates change. Everything on this page was read from Nanonets' own pricing page in July 2026, and we would rather you verify it there than trust us.
Read from Nanonets' own pricing page in July 2026. Where a number is not published, this table says so rather than guessing.
| Line item | Published price | What it covers |
|---|---|---|
| Simple operations | $0.02 per block run | Formatting a field, routing, basic steps |
| Standard AI | $0.10 per block run | Ordinary AI extraction and classification steps |
| Complex AI | $0.30 per block run | The heavier AI blocks Nanonets quotes in its own invoice example |
| Starter plan | $50 in credits to start, then $100 a month for 100 credits | Data extraction, API access, email integration, cloud storage, up to 3 users |
| Growth plan | Quote only, up to 40% volume discount | Classification, barcode and signature detection, ERP and database integrations |
| Enterprise plan | Custom, tailored to your volume | SAML SSO, SCIM, RBAC, HIPAA and SOC 2, private cloud or on-prem, audit logs |
| Credit expiry | Credits never expire, shared across the team | Suits bursty workloads such as a quarter-end close |
One line on that table is missing a number, and it is the one you need most. Nanonets sells credits at $100 for 100, describes them as prepaid usage spendable across any workflow block, and separately prices those blocks in dollars. Nowhere does it state how many dollars of block runs one credit buys. Until you have that ratio in writing, any Nanonets estimate you build is an estimate of an estimate. Credits never expiring is a genuine plus, and it suits workloads that arrive in bursts rather than evenly across a month.
Published rates as of July 2026, taken from each vendor's own pricing page. AWS figures are US West (Oregon). The row that matters most is the fourth one, because it explains why the third one looks so lopsided.
| Dimension | Nanonets | Azure AI Document Intelligence | AWS Textract | Google Document AI | DocuOCR |
|---|---|---|---|---|---|
| Billing unit | Block run | Page | Page | Page | Page |
| Published headline rate | $0.02 to $0.30 per block run | About $30 per 1,000 pages (custom) | About $50 to $70 per 1,000 (Forms) | About $30 per 1,000 (Form Parser) | About $14 to $20 per 1,000 pages |
| Cost of one 1-page invoice | About $1.20 to $1.80, per Nanonets own example | About $0.03 | About $0.05 to $0.07 | About $0.03 | About $0.014 to $0.02 |
| What that price includes | Classify, extract, format, validate, post to ERP | One model call returning fields | One model call returning fields | One model call returning fields | Classify, extract, validate, review, export |
| Steps you must build yourself | None of the above | All of the above | All of the above | All of the above | None of the above |
| All tiers publish a price | No, 2 of 4 are quote only | Yes | Yes | Yes | Yes |
Rows three and four have to be read together or not at all. The cloud APIs look almost free next to Nanonets because they are selling one step of a five-step job. Nanonets is selling all five. Whether that is worth the difference depends entirely on whether you have engineers who want to build and then maintain the other four, and on how often your document formats change. For the full cross-vendor picture, see our OCR API pricing comparison. If you want the cheapest raw reader in the market, that is currently Gemini OCR pricing, at roughly $0.33 per 1,000 pages, and it gives you nothing but text.
A realistic US mid-market accounts payable workload: 5,000 single-page invoices a month that have to end up as validated records in an ERP, not as text on a screen.
$6,000 to $9,000
per month
Nanonets own example: 4 to 6 complex blocks at $0.30. The whole workflow, posted to your ERP.
$4,500
per month
Two complex plus three standard runs is $0.90 a document. Design the workflow with cheaper blocks and the bill moves.
$150
per month
About $30 per 1,000 pages for the extraction call alone. Classification, validation, review, and ERP posting are yours to build.
$149
per month
The published 10,000 page plan, with classification, validation, review, and export included.
The Azure column is not really $150. It is $150 plus the engineers who build classification, write the validation rules, ship a review screen, and maintain the ERP integration, and then keep doing it as vendors change their invoice layouts. For a lot of US mid-market teams that work costs more in year one than any of these line items. The Nanonets columns are honest about including it. The gap between the two Nanonets columns is worth noticing on its own: the same document through a workflow designed with standard rather than complex blocks costs about half as much, so how you build the workflow matters as much as which vendor you pick. Price the whole job, then compare.
Azure, AWS Textract and Google Document AI per 1,000 pages, side by side.
About $0.33 per 1,000 pages, and why output tokens are the whole bill.
$4 per 1,000 pages for OCR 4, and why the $1 figure is stale.
$1.50 per 1,000 units, and why Vision returns no form fields.
The other platform that gates its tiers: $18,000 a year to start.
What changes when the workflow is priced per page instead of per block.
How the platform works and what it returns.
S0 rates for Read, Layout, prebuilt and custom extraction.
The workflow those block runs are actually performing.
An honest roundup of the platforms for US teams.
A rate card cannot tell you whether the fields come out right. Upload one of your real documents, look at what comes back, and then decide what unit you want to be billed in.
