# PennyOCR

> OCR API for developers and agents: POST a document (PDF/PNG/JPEG/WebP/TIFF),
> get its full text back as markdown (tables preserved) or plain text.
> $0.75 per 1,000 pages, first 100 pages free. Zero data retention by default.

## MCP server (recommended for agents)

Add to mcpServers: {"pennyocr": {"url": "https://mcp.pennyocr.com",
"headers": {"Authorization": "Bearer pk_live_..."}}}
Tools: read_document {url, pages?, output?, max_pages?, max_cost_usd?} ->
{content, pages, cost_usd, citations[]} · estimate_cost {url, pages?} (free).
Clients without header support: https://mcp.pennyocr.com/?key=pk_live_...

## API (base: https://api.pennyocr.com)

- POST /v1/ocr — multipart form, field `file`; query `?format=markdown|text`;
  header `Authorization: Bearer pk_live_...`. Response JSON: {id, pages, text,
  page_results[], cost_usd, model}. Errors: 401 bad key, 402 out of credits,
  413 too large (50MB/500 pages), 422 unreadable.
- POST /v1/chat/completions — OpenAI Chat Completions compatible. Send image_url
  content parts (data: or https URLs); model "pennyocr-markdown" or "pennyocr-text".
- GET /v1/models — OpenAI-style model list.
- POST /v1/ocr/url — JSON {url, format?, pages?, max_pages?, max_cost_usd?};
  fetches the document server-side. 413 (with exact numbers) if a cap would
  be exceeded — decide, then retry.
- POST /v1/estimate — JSON {url, pages?} -> {total_pages, selected_pages,
  cost_usd}. Free.
- GET /v1/balance — (Authorization: Bearer pk_live_...) your key's budget:
  {credit_pages, credit_value_usd, free_pages_available_now, price_per_1k_usd}.
  Free. Check this before spending; the first 100 pages on an account are free.
- POST /v1/billing/checkout — (Authorization: Bearer pk_live_...) JSON
  {pack: "10k"|"100k"|"1m"} -> {url} to buy prepaid credits with just the API
  key (no session token). A human finishes payment in the browser.
- GET /health — status, no auth.
- Machine-readable spec: https://api.pennyocr.com/openapi.yaml (and /openapi.json)

## Signup for agents (no browser needed)

1. POST https://api.pennyocr.com/auth/request-code    {"email": "you@example.com"}
2. Read the 6-digit code from that inbox, then
   POST https://api.pennyocr.com/auth/verify          {"email": "...", "code": "123456"}
   -> {"session_token": "sk_session_..."}
3. POST https://api.pennyocr.com/dashboard/keys       (Authorization: Bearer <session_token>)
   -> {"api_key": "pk_live_..."}  (shown once)
4. Your first 100 pages are free. To buy prepaid credits:
   POST https://api.pennyocr.com/dashboard/billing/checkout
   {"pack": "10k" | "100k" | "1m"}  (Authorization: Bearer <session_token>)
   -> {"url": "https://checkout.stripe.com/..."}  (complete payment there; a
   human card holder must finish Stripe checkout)
   Or, once you hold the API key: POST /v1/billing/checkout {"pack": "..."}.
5. Balance (with the API key): GET /v1/balance. Session usage: GET
   /dashboard/usage · daily series: GET /dashboard/usage/daily

## Pricing

- /pricing.txt: plain-text pricing summary
- https://pennyocr.com/ocr-api-pricing/: pricing comparison page

## Docs

- https://pennyocr.com/docs/: full API reference
- https://pennyocr.com/blog/receipt-ocr-benchmark/: measured accuracy benchmark

## Contact

- hello@pennyocr.com (a human replies fast)

## Chrome extension
PennyOCR Snip — snip any region of a page (or right-click an image) and get the text.
Install: https://chromewebstore.google.com/detail/pennyocr-snip/mpbmgjggnjmembmagibkgeldehgcipgn
Source: https://github.com/imiraoui/pennyocr-snip
3 free snips/month with no account; email sign-in raises it to 30/month.