Hardparsebeta
Hardparse OCR

Drop a document. Get the parsed result.

Upload one PDF, image, screenshot, or WebM. Hardparse returns Markdown, reading order, confidence, and source boxes.

No signup, no email, no extra form. Use the big upload box below.

Good forRAG, invoices, scans, QA review
ReturnsMarkdown + layout bbox JSON
StatusPublic beta is live
01

Layout

Detects document regions, categories, confidence, and reading order before turning the page into text.

02

Bboxes

Switch between layout blocks and text-line overlays, then click any box to inspect its source.

03

Formats

PDFs, scans, screenshots, WebP, and WebM first frames run through the same live parser.

One upload endpoint.

The public beta accepts anonymous uploads with rate limits. For higher volume or private deployments, email k.wikiel@gmail.com.

curlpythonnode
POST /v1/parse
# Public beta: no key required for low-volume testing
curl https://hardparse.com/v1/parse \
  -F file=@document.pdf \
  -o result.json

jq '.markdown' result.json
jq '.pages[0].regions[0].box' result.json
Ship it

Need OCR inside your product?

Try the public beta first. If you need a private instance, higher limits, or integration work, email k.wikiel@gmail.com.

Email me