Nanonets pricing is usage based: every account starts with $50 in credits, the self-serve Starter plan is $100 a month for 100 credits, and every step of a document workflow draws down credits separately. AI data extraction costs 0.30 credits per page, simple formatting 0.02 per run, and exports and lookups 0.04 to 0.15. There are no seat licenses and no platform fee. Growth and Enterprise are quote based, with volume discounts of up to 40%.
The headline rates are easy to find. What is hard to find is how many times each rate gets charged on one document, and that is where a Nanonets bill is decided. The figures below come from the Nanonets pricing page and its consumption billing documentation, not from a directory listing, and several of the directory listings are out of date (more on that below).
How much does Nanonets cost per month?
There are three plans. Only Starter has a published price.
| Plan | Price | What it adds |
|---|---|---|
| Starter | $50 in starting credits, then $100 a month for 100 credits | Data extraction AI, API access, email import, cloud storage connectors, up to 3 users, community support |
| Growth | Quote, volume discounts up to 40% | Classification AI, barcode and signature detection, generative AI blocks, custom Python blocks, ERP and database integrations, analytics, team-wide credit sharing |
| Enterprise | Custom | SAML SSO and SCIM, role-based access, HIPAA and SOC 2 compliance, private cloud or on-prem, data residency, Salesforce, SAP and Oracle connectors, SLAs, audit logs |
At the Starter list price a credit costs one dollar, which makes the rest of the arithmetic simple: a rate of 0.30 credits is 30 cents.
How does Nanonets charge per document?
It does not charge per document at all. It charges per block run. A workflow is a chain of blocks (import, extract, format, look up, validate, export) and each block has its own price and its own rule for what counts as one run. This is the rate table from the Nanonets documentation.
| Block | Credits per run | What counts as one run | Availability |
|---|---|---|---|
| Import (email, API, Google Drive, Dropbox, SharePoint, OneDrive) | 0.04 | Each file imported | All plans |
| Data Extraction AI | 0.30 | Each page processed | All plans |
| Classification, checkbox, barcode, signature or QR detection AI | 0.10 | Each run | All plans |
| Data formatting (dates, numbers, text, regex) | 0.02 | Each field occurrence, or each table row per column | All plans |
| Python block or LLM post-processing | 0.18 | Each run | All plans |
| Lookup: Google Sheets, databases, CSV | 0.04 | Per document, or per row when matched against rows | All plans |
| Lookup or export: QuickBooks, Sage, Xero, API | 0.10 | Export counts each page processed | All plans |
| Lookup or export: Salesforce, SAP, Oracle, NetSuite, D365 | 0.15 | As above | Enterprise |
Two rows in that table do most of the damage, and neither one is extraction.
Why a Nanonets bill grows with table rows
Formatting is the cheapest block on the list, at 0.02 credits. It is also the only one that multiplies by the number of rows. The documentation is explicit: formatting a single column of a ten row table is ten runs, and formatting two columns of that table is twenty. Setting up a date formatter on five fields that each appear three times is fifteen runs.
On a one page invoice with a dozen line items that is a rounding error. On the documents that finance teams actually struggle to get into a spreadsheet, it is the biggest line on the bill. Take a twelve page supplier statement with 360 transactions, and format the date and amount columns so they land as real dates and real numbers: that is 720 formatting runs, or 14.40 credits. Extracting all twelve pages cost 3.60. The cheapest block in the workflow cost four times as much as the AI.
Export works per page too, so a 12 page file pushed to QuickBooks is 12 export runs at 0.10, even if it creates a single record.
What does one document actually cost on Nanonets?
Here is the same rate table applied to five realistic documents. Credits convert to dollars at the Starter list price of $1.
| Document and workflow | Runs charged | Credits |
|---|---|---|
| 1 page invoice, 10 line items, extract, export to QuickBooks | Import 1, extract 1, export 1 | 0.44 |
| 1 page invoice, 20 line items, format the amount column, export to QuickBooks | Import 1, extract 1, format 20, export 1 | 0.84 |
| 3 page invoice, 60 line items, format date and amount, export to QuickBooks | Import 1, extract 3, format 120, export 3 | 3.64 |
| 12 page statement, 360 rows, extract and export to Google Drive only | Import 1, extract 12, export 12 | 4.12 |
| 12 page statement, 360 rows, format date and amount, export to Google Drive | Import 1, extract 12, format 720, export 12 | 18.52 |
Nanonets says a typical invoice workflow runs four to six blocks and comes in under $2 per invoice. The first two rows agree with that. The last three show why "per invoice" is the wrong unit for long, table heavy documents: page count multiplies extraction and export, and row count multiplies formatting.
For budgeting, the $100 Starter month covers roughly 227 of the simplest invoices in that table, about 119 of the second kind, and five of the formatted statements.
What costs extra on top of block runs?
Some features are billed as a flat monthly add-on rather than per run, and they are easy to miss because they are not on the pricing page itself. From the Nanonets rate documentation:
| Add-on | Credits per month | Availability |
|---|---|---|
| AI confidence scores | 500 | All plans |
| SAML SSO | 500 | Enterprise |
| Analytics package | 200 | Enterprise |
| Whitelabel UI | 200 | Enterprise |
| Audit and file management package | 100 | Enterprise |
| UAT model | 100 | Enterprise |
| User access levels | 50 per user | Enterprise |
| Human in the loop review (managed service) | 1.00 per page | Enterprise |
The one to notice is confidence scores. A confidence score is how a reviewer knows which extracted values to double check, and it is listed at 500 credits a month, which is five times the Starter plan's own monthly spend. A team processing 500 of the 0.84 credit invoices above would pay about $420 in runs and $500 for the scores.
Security is the other place the tier matters. HIPAA and SOC 2 compliance sit on Enterprise, so if a vendor attestation has to go into your own audit file, the people who keep your regulatory monitoring current will need the Enterprise conversation to happen before anyone signs.
Is Nanonets free?
There is no ongoing free plan. Every account starts with $50 in credits with no card required, which at the rates above is enough to run a realistic pilot of 60 to 110 simple invoices, and after that usage is paid. Nanonets also notes that failed block retries are not charged, and that there are no platform fees or monthly minimums.
Why do G2 and Capterra show different Nanonets prices?
Because they describe the pricing Nanonets used before it moved to block runs. Directory listings still quote editions from $0 to $999 a month and a $499 Pro plan, and some reviews quote "about $0.30 per page", which is only the extraction block. The Nanonets pricing page itself tells existing customers on legacy pricing to contact their account manager, which confirms the older plans still exist for some accounts but are not what a new buyer gets. Its own documentation is not perfectly consistent either: the worked examples in the billing guide use an import rate of 0.01 and an export rate of 0.03, while the rate table lists 0.04 and 0.10. The figures in this article use the rate table, which is the more expensive reading. Credits on the pricing page are described as never expiring, while the billing guide says expiry terms can be negotiated with sales, so confirm the term in your order form.
Is Nanonets worth it for PDF to Excel?
It depends on what happens after extraction. Nanonets is built to run a process: classify a document, pull named fields, route exceptions to an approver, and post the result into QuickBooks, NetSuite or SAP. If that whole chain is the job, block pricing is a fair way to pay for it, and the ERP connectors, approvals and compliance options are things a spreadsheet tool simply does not have.
If the job ends at a spreadsheet, you are paying for a workflow engine to do a conversion, and the per-row formatting charge lands exactly on the part you care about, which is getting clean numbers and dates into columns. That is the case our PDF parsing tool is built for. It returns every table as rows in XLSX or CSV, reads scans with OCR on every plan, and counts one conversion per file however many pages or rows it has. Plans are flat: Starter at $9 a month for 50 files, Plus at $24 for 500 with batch upload and saved column mappings, and Pro at $49 for unlimited files and the PDF to Excel API. The full comparison is on our pricing page.
| Question | Nanonets | PDFxlsx |
|---|---|---|
| Billing unit | Each block run: per page, per row, per field | Per file |
| 12 page, 360 row statement with clean dates and amounts | 18.52 credits | 1 conversion |
| Output | Named fields routed through a workflow | Every table as rows in XLSX, CSV or JSON |
| Approvals, ERP posting, HIPAA | Yes, Growth or Enterprise | No |
| Scanned PDFs | Yes | Yes, OCR on every plan |
| Published entry price | $100 a month for 100 credits | $9 a month for 50 files |
If you are still choosing between tools rather than costing this one, the Nanonets alternative page compares the two feature by feature, and converting invoice PDFs to Excel shows the invoice tie-out we recommend running on whatever tool you pick.
How do Nanonets costs compare with Parseur and Docparser?
They meter different things, which is why a like-for-like price comparison rarely works. Parseur pricing counts one credit per PDF page, so long documents are expensive and row count does not matter. Docparser pricing counts documents, so many short files are expensive and page count matters less. Nanonets counts pages for extraction and export and rows for formatting, so it is the one model where a long, dense table costs you twice. Before you compare quotes, write down three numbers for a typical month: files, average pages per file, and average table rows per file. Every vendor's price becomes comparable once you have those.