* * * PADDLEOCR, HOSTED VS DIY * * *

PaddleOCR as an API.

PaddleOCR is genuinely great open-source OCR. But "free" means: a GPU box, CUDA drivers, a serving layer, queueing, monitoring, and you on call when it OOMs at 2am. The build-vs-buy math: a hosted endpoint at $0.75 per 1,000 pages vs your own PaddleOCR deployment.

START FREE — 50 PAGES
$0.75 / 1,000 PAGES
1,333 PAGES PER DOLLAR · NO CREDIT CARD
AT A GLANCE
UPDATED
PRICE
$0.75 per 1,000 pages ($0.00075 per page), prepaid
FREE TIER
First 50 pages free per account, no credit card
FORMATS
PDF, PNG, JPEG, WebP, TIFF
LIMITS
50 MB and 500 pages per request
OUTPUT
JSON with markdown (tables preserved) or plain text, per page and joined
INTERFACES
REST API, OpenAI-compatible chat completions, MCP server
CREDITS
Never expire; requests stop with a 402 when the balance runs out
01 / TRY IT

ONE ENDPOINT.

POST a file, get JSON back — the extracted text, per page and joined. PDF, PNG, JPEG, WebP or TIFF.

curl https://api.pennyocr.com/v1/ocr \
  -H "Authorization: Bearer $PENNYOCR_API_KEY" \
  -F "file=@scan.jpg"

# $0.75 per 1,000 pages, first 50 free
02 / USE CASES

THE REAL COSTS OF DIY.

THE GPU BILL
An always-on cloud GPU runs $150–500+/month idle or not. That's 200,000–650,000 PennyOCR pages.
THE OPS TAX
Serving, batching, retries, upgrades, model updates — engineering days that don't ship your product.
WHEN DIY WINS
Hard data-locality/airgap requirements, massive steady volume (10M+ pages/month), or OCR IS your product. Then self-host — maybe start from PaddleOCR-VL.
THE HOSTED DEAL
One endpoint, markdown with tables, measured accuracy, zero retention by default, pay per page. No pager.
03 / PRICE CHECK

HALF THE PRICE OF THE BIG CLOUDS.

Per 1,000 pages, public list prices, first tier.

PENNYOCR$0.75
AWS TEXTRACT$1.50
GOOGLE CLOUD VISION$1.50
AZURE DOC INTELLIGENCE$1.50
YOU KEEP50%
04 / FAIR QUESTIONS
GET YOUR API KEY →
QUESTIONS? HELLO@PENNYOCR.COM
IS PENNYOCR JUST HOSTED PADDLEOCR?
No — we serve a modern vision-language model (doc-to-markdown), a different architecture than PaddleOCR's detector/recognizer pipeline. Different strengths; VLMs handle messy layout and handwriting better.
AT WHAT VOLUME DOES SELF-HOSTING WIN?
Rough break-even against a $300/month GPU + your time is 400k–1M+ pages/month, sustained. Below that, hosted wins on cost alone before counting engineering time.
CAN I MIGRATE LATER?
Trivially — it's OCR, not lock-in. Start hosted, and if you outgrow us, move to self-hosting anytime. We'll still be cheaper than your GPU bill for most volumes.