PDF OCR API

Extract text from PDFs via a simple HTTP API. Text-native pages use fast parse; image-heavy / scanned / handwritten pages use PaddleOCR-VL-1.6 via hosted worker (capped). Auth with a Better Call API key.

Step 1

Get an API key

Keys are shown once. We store a SHA-256 hash only.

Email

Step 2

Usage today

Paste a key to check remaining free-tier quota (30 / day).

API key

— / —

Try

OCR a PDF (browser)

Calls the live function with your key. Max 10 MB, 10 pages.

API key PDF file

Docs

Auth & endpoints

Authorization: Bearer bc_pdf_...

Method	Path	Notes
POST	`/v1/keys`	body `{email}` → key once
GET	`/v1/keys/me`	metadata, no secret
POST	`/v1/ocr`	JSON `file_base64` or multipart
POST	`/v1/billing/checkout`	`pdf_key\|email` + `price_id`
POST	`/v1/billing/portal`	`pdf_key\|customer_id`
POST	`/v1/billing/webhook`	Stripe signature

Curl (Appwrite executions)

Function URL

Site JS posts to the Appwrite function executions endpoint. Direct HTTP domain for the function is optional.

POST https://nyc.cloud.appwrite.io/v1/functions/pdf-ocr/executions
X-Appwrite-Project: 69e5a408000ff43aa282
Content-Type: application/json

Limits

MVP caps

30 OCR requests / day / key (configurable)
Max upload 10 MB · max 10 pages OCR'd
Text extract first; OCR if text is sparse
No long-term PDF storage