Glint reads invoices, receipts, contracts and forms, then returns typed fields from one API call. Dates in ISO 8601, money as numbers, and a warning when the math does not add up.
Text layer or OCR. Every word gets a position on the page.
The page is cut into tokens: words, numbers, dates, amounts.
Tokens land on fields in your schema. Labels and noise are dropped.
Line items must sum to the subtotal. Dates come back as ISO 8601.
Text layer or OCR. Every word gets a position on the page.
Northfield Print Co.
418 Mill St, Portland OR
Invoice
No. INV-2048
| Description | Qty | Unit | Amount |
|---|---|---|---|
| Brand guidelines | 1 | $1,850.00 | $1,850.00 |
| Business cards | 500 | $0.42 | $210.00 |
| Letterpress setup | 2 | $95.00 | $190.00 |
Payment by bank transfer within 30 days. All amounts in USD.
{ "type": "invoice", "number": "INV-2048", "issued": "2026-03-14", "due": "2026-04-13", "currency": "USD", "vendor": { "name": "Northfield Print Co.", "address": "418 Mill St, Portland OR" }, "bill_to": "Harbor & Vale Ltd", "line_items": [ { "desc": "Brand guidelines", "qty": 1, "unit": 1850.00, "amount": 1850.00 }, { "desc": "Business cards", "qty": 500, "unit": 0.42, "amount": 210.00 }, { "desc": "Letterpress setup", "qty": 2, "unit": 95.00, "amount": 190.00 } ], "subtotal": 2250.00, "tax": 180.00, "total": 2430.00}Change any field. Write a date as 14/03/2026, a price in euros, a quantity as a word. The parser on this page runs in your browser so you can poke at the idea. The API does the same job on real files.
Post a file with the schema you want. The response has the shape you just watched: named fields, numbers you can add up, and a warnings array instead of silent guesses.
Authorization header.invoice, receipt, bank_statement, or your own JSON Schema.curl https://api.glint.example/v1/extract \
-H "Authorization: Bearer $GLINT_API_KEY" \
-F "file=@invoice.pdf" \
-F "schema=invoice"import os, requests
resp = requests.post(
"https://api.glint.example/v1/extract",
headers={"Authorization": f"Bearer {os.environ['GLINT_API_KEY']}"},
files={"file": open("invoice.pdf", "rb")},
data={"schema": "invoice"},
)
invoice = resp.json()
print(invoice["total"], invoice["currency"])import { readFile } from "node:fs/promises";
const form = new FormData();
form.append("file", new Blob([await readFile("invoice.pdf")]), "invoice.pdf");
form.append("schema", "invoice");
const res = await fetch("https://api.glint.example/v1/extract", {
method: "POST",
headers: { Authorization: `Bearer ${process.env.GLINT_API_KEY}` },
body: form,
});
const invoice = await res.json();
console.log(invoice.total, invoice.currency);The endpoint shown is a placeholder for this concept.
Send the file you already have. Glint picks the reader for it, and every format comes back as the same JSON.
You pay for pages read. A three-page PDF counts as three. Every plan gets every format, custom schemas and webhooks.
| Plan | Per 1,000 pages | What it covers |
|---|---|---|
| Sandbox | $0 | The first 500 pages every month.No card. Same API, same output. |
| Build | $12 | Pages 501 to 250,000 in a month.Pay as you go, billed monthly. |
| Scale | $8 | Every page past 250,000 in a month.Applied automatically, no contract. |
Prices are part of this concept and would be set by the product team before launch.
Manual cost is pages times minutes per page, divided by 60, times your hourly rate. The 4-minute default is a sample; check it against your own team's timing.
500 pages a month are free, every month. No card needed.