1099 OCR APIs run from about half a cent to three cents per form. Mistral Document AI is the cheapest verified structured rate at $0.005, Azure's prebuilt US Tax 1099 model about $0.01 per page, and the Google or AWS custom-extraction fallback about $0.03 because neither ships a 1099 parser.
Azure is the only major cloud with a prebuilt 1099 model, and it covers 1099-NEC, MISC, DIV, INT and eighteen more variants. Here is the honest math. Last updated July 2026.
Upload a document to extract
Drop files here or click to upload
Up to 50 files
Uploading...
Before you wire up an API, drop in a real 1099-NEC or 1099-MISC and see the payer TIN, recipient TIN and the numbered boxes come back as clean fields.
A 1099 OCR API costs between about half a cent and three cents per form for the box values, and the number depends less on the billing unit than on whether the vendor has a 1099 model at all. A 1099 is a single page, so per-page and per-document units nearly converge. Mistral Document AI bills per page at $0.005, the cheapest verified structured rate, on a generic schema you map to the boxes. Azure's prebuilt US Tax 1099 model bills per page at $0.01 and is the only major-cloud model that reads the base 1099 plus about twenty variants directly, from 1099-NEC and 1099-MISC to 1099-DIV, 1099-INT and 1099-R. Google Document AI ships a W2 parser and a Pay Slip parser but no 1099 processor, and AWS Textract has none either, so on both you fall to a custom extraction model at roughly $0.03 a page, three times Azure, or you map plain OCR yourself. For a 1099-native schema, Azure is both the cheapest dedicated path and the most complete, which is unusual.
A 1099 looks like a W-2 pricing problem, one page, box values, but it is not, because of model coverage. For a W-2 you can pick Azure's per-page prebuilt model or Google's dedicated W2 parser. For a 1099 that choice narrows to one. As of July 2026 Azure AI Document Intelligence is the only major cloud with a prebuilt 1099 model, and it reads the base 1099 plus about twenty variants: A, B, C, CAP, DIV, G, H, INT, K, LS, LTC, MISC, NEC, OID, PATR, Q, QA, R, S, SA and SB. That includes 1099-NEC for contractor pay and 1099-MISC for miscellaneous income, the two a business handles most.
Google Document AI ships a W2 parser and a Pay Slip parser, but no 1099 processor. AWS Textract has no 1099 model either. On both you either run a custom extraction model at about $30 per 1,000 pages, three times Azure's prebuilt 1099 rate, or you read the form with plain OCR at $1.50 per 1,000 and map the payer TIN, recipient TIN and each box yourself. Mistral reads a 1099 at $0.005 a page on a generic schema, cheaper still, but again you own the box mapping and the variant differences.
The practical rule for 1099s: if you want a ready 1099 schema across many variants, Azure's prebuilt model is the cheapest and most complete dedicated path. If you are fine mapping a generic schema once and reusing it, Mistral is cheaper per page. Only fall to Google or AWS custom extraction if you are already committed to that cloud for other reasons.
Rates and model coverage change. Everything on this page was read from each vendor's own pricing and model docs in July 2026, and we would rather you confirm it there than trust us.
Read from each vendor's own pricing page in July 2026 and normalized to a single-page 1099. Where a vendor has no dedicated 1099 model, this table says so.
| Vendor | Cost per 1099 | Billing unit | 1099 model? | Free allowance |
|---|---|---|---|---|
| Mistral Document AI | $0.005 per 1099 | Per page ($5 per 1,000) | Generic, you map to boxes | No standing free tier |
| Azure AI Document Intelligence (US Tax 1099) | About $0.01 per 1099 | Per page ($10 per 1,000) | Yes, ~20 1099 variants | 500 pages a month (F0) |
| DocuOCR | About $0.014 to $0.02 per 1099 | Per page (product) | Yes, plus review and export | On your own documents |
| Google Document AI | About $0.03 per 1099 | Per page ($30 per 1,000, Custom Extractor) | No 1099 model, build it | New-customer credit |
| AWS Textract | About $0.05 to $0.07, you build it | Per page ($50 to $70 per 1,000, Queries/Forms) | No 1099 model, build it | 100 pages a month, 3 months |
| Veryfi | Not published for 1099 | Per document (W-2/W-9 are $0.16) | W-2/W-9 yes, 1099 custom | 100 documents a month |
Three rows on that card need your own numbers. Google and AWS Textract have no dedicated 1099 model, so those figures are what it costs to run a custom extraction model or write Queries and map the boxes yourself. The DocuOCR figure is a product range that includes classification, review, and export, not a raw OCR rate. For the full per-vendor rate cards, see the OCR API pricing comparison.
A tax season volume of single-page 1099s run through each option, so the model gap and the custom-extraction premium show up in dollars.
| Option | Monthly cost | Why |
|---|---|---|
| Mistral Document AI | $100 | 20,000 1099s at $0.005 each. Cheapest verified rate, generic schema you map. |
| Azure (US Tax 1099 model) | $200 | 20,000 1099s at $0.01 on the prebuilt 1099 model, ~20 variants, per page. |
| DocuOCR | $280 to $400 | A product that also classifies, reviews low-confidence boxes, and exports. |
| Google / AWS custom extraction | $600 | No 1099 model, so a custom extractor at about $0.03 a page. |
Six times the spread, from $100 to $600, for the exact same 20,000 1099s, decided by the billing unit and by who has a 1099 model. Azure's prebuilt 1099 model beats the Google or AWS custom-extraction fallback three to one because it prices per page, not as a custom model. Mistral is cheaper still if you map a generic schema. A ready-to-use product like DocuOCR sits in the middle and adds the review step that matters when a wrong recipient TIN becomes a wrong information return. See the per-vendor detail on W-2 OCR API pricing and Azure Document Intelligence pricing.
The sibling one-page form, where Google also has a dedicated parser.
Azure, AWS Textract and Google Document AI per 1,000 pages, side by side.
The per-receipt cost, where a single page flips the math the same way.
Multi-page statements, where the per-document parsers flatten the cost.
$4 to $5 per 1,000 pages, the cheapest structured extraction rate.
Turn 1099-NEC, MISC, DIV and INT forms into mapped fields.
Classify a client stack and extract W-2s, 1099s, K-1s and 1098s.
An honest roundup of the extraction APIs for US teams.
Which vendors have prebuilt W-2, 1099 and pay stub models.
A per-form rate cannot tell you whether the payer TIN, recipient TIN and the numbered boxes come out right on your real 1099-NEC and 1099-MISC forms, including the odd faxed or photographed one. Upload a form, look at the mapped boxes, and then decide which billing unit you want.