Send a PDF. Get clean Markdown your LLM can read.
One API call. Tables preserved. Headings preserved. Footnotes inline. Half the price of LlamaParse.
PDF parsing is broken.
Most AI apps that “chat with documents” feed garbage to the model. Tables get scrambled. Multi-column pages mix together. Footnotes inline with body text. Then the LLM hallucinates and the user blames the AI.
Mintmark gives you the parsing layer that actually works — backed by two best-in-class open-source projects (Microsoft markitdown and IBM docling), wrapped in a single REST endpoint, billed per page.
One call
From PDF to Markdown in one request.
Request
curl -X POST https://api.mintmark.dev/v1/convert \
-H "Authorization: Bearer $MINTMARK_KEY" \
-F "file=@contract.pdf" \
-F "mode=balanced"Response
{
"id": "req_01HX...",
"status": "succeeded",
"data": {
"markdown": "# Section 12 — Indemnification\n\n12.3 The Party of First Part shall...\n\n| Term | Duration | Penalty |\n|------|----------|---------|\n| Non-compete | 2 years | $50,000 |",
"metadata": {
"engine_used": "markitdown",
"page_count": 42,
"tables_extracted": 4
}
},
"usage": { "units": 42, "unit_type": "page" }
}Receipts
Same quality. 10× cheaper.
Numbers from public pricing pages of each competitor as of May 2026.
| LlamaParse | Unstructured | Reducto | Mintmark | |
|---|---|---|---|---|
| Per-page price | $0.30 | $0.10 | $0.10–$0.20 | $0.02 |
| Setup | SDK + auth | Python SDK | Sales call | One curl |
| Tables preserved | Yes | Yes | Yes | Yes |
| Self-serve checkout | Yes | Yes | No | Yes |
| OSS-backed | Closed | Open-core | Closed | MIT (markitdown + docling) |
| Free tier | 1,000 pages/day | Limited | None | 100 pages/mo |
Pricing
Simple, usage-based pricing.
Start free. Upgrade when you need more. No annual contracts.
Scale
$499/ month
250,000 pages
- Premium mode included
- SLA
- Dedicated support
- Self-host enterprise option
Want early access to enterprise self-host?
Drop your email. We’ll reach out when SOC 2 closes.