Convert PDF to Markdown and spend up to 80% less on AI credits
An LLM reads each PDF page as extracted text plus an image: about 2,300 tokens. The same content as clean Markdown costs a fraction of that, and the context gets more useful.
Example: a 40-page report sent to Claude
40 pages: 72% fewer input tokens for the same question.
Drop your document here
PDF, Word, Excel, CSV, HTML and more · up to 10 MB · up to 100 pages
No sign-up for your first conversions · The file is discarded when done
No sign-up to start
3 conversions per day, right in the browser.
Nothing is stored
Direct conversion in memory, discarded with the response. In the queue: file deleted on completion, result expires in 24 h.
Scanned PDF? OCR
On paid plans, text is recognized page by page.
Seconds per document
Long documents go to a queue and you follow the progress.
Markdown ready for
- ChatGPT
- Claude
- Gemini
- Cursor
- Notion
- Obsidian
How it works
Three steps, no tutorial
You already know how: drop the file, grab the text, paste it into the conversation with the model.
Upload the document
Drag the file or pick it from your computer. PDF, Word, Excel, CSV, HTML and, on paid plans, PowerPoint and EPUB.
- DOCX
- XLSX
- CSV
- HTML
- PPTX
- EPUB
Get the Markdown
Headings, lists and tables preserved; repeated headers and footers removed. You see the savings right away.
Paste into your LLM
Copy or download the .md and use it in Claude, ChatGPT, Gemini, Cursor or your own agent. Fewer tokens, cleaner context.
Before and after
The same document, now readable by the model
The PDF comes in with broken columns and a table turned into loose text. The Markdown comes out with headings, table and list in place.
Scrambled columns, the table became loose text, footer repeated on every page.
≈ 9,200 tokens
# Annual report 2025
## Highlights
Net revenue grew 12% and gross
margin reached 41%.
| Quarter | Revenue | Margin |
| ------- | ------- | ------ |
| Q1 | 2.1M | 38% |
| Q2 | 2.4M | 40% |
| Q3 | 2.6M | 42% |
| Q4 | 2.9M | 43% |
## Risks
- FX
- Customer concentration
Headings, table and list are still headings, table and list.
≈ 1,400 tokens
The math
Why the same document costs so much less as Markdown
The converter doesn’t compress or summarize: it removes what the model doesn’t need to read, such as the page image and repeated headers.
Page as PDF
~2,300 tokens
Each page becomes extracted text plus a rendered image the model also has to "read".
Page as Markdown
~300 to 800 tokens
Only the content: headings, tables and lists. No image, no repeated header and footer.
Result
50 to 80% savings
Fewer credits spent, more precise answers and none of the page limits of the PDF pipeline.
Calculator
How many tokens do you burn every month?
Drag to your monthly page volume and see the difference.
500
PDF pages per month
You stop spending
825,000 tokens
In dollars, per month
$4
72% less, on input tokens alone with Claude Opus 5 (US$ 5 per million).
Estimate. A PDF reaches the model as extracted text plus the page image (~2,300 tokens); the same content as plain Markdown is close to 650. The real savings of each file show on screen when you convert.
Who it’s for
Built for people who pay per token
If your work involves sending long documents to a model, the cost of PDF shows up on the invoice.
Developers and AI agents
Feed RAG, agents and pipelines with clean Markdown instead of raw PDF. Fewer tokens per call, less noise in context.
Law and legal teams
Contracts, filings and rulings hundreds of pages long fit in context without blowing the plan. OCR for what arrived scanned.
Research and teaching
Papers, theses and course material become structured text to summarize, compare and study with the model.
Finance and reporting
Statements, balance sheets and annual reports with tables preserved, ready for analysis.
Plans
Start free and upgrade when you need more
Every plan uses the same engine. It runs on any system, nothing to install, and OCR is on our servers. What changes is the volume, file size, batch size and OCR access.
Free
R$0
Free forever, no card
To really try it: 30 conversions a month, no card.
- 30 conversions per month
- Up to 10 MB and 100 pages per file
- PDF, Word, Excel, CSV and HTML
- Up to 3 files at a time
- No OCR for scanned PDFs
- Direct conversion only (up to 100 pages)
Pro
R$29/month
Billed monthly, cancel anytime
For those who send long reports and contracts to the model every week.
- 500 conversions per month
- Up to 50 MB and 1,000 pages per file
- All formats, including PowerPoint and EPUB
- Up to 10 files at a time
- 50 OCR conversions per month
- Queue for long documents, with progress
Ultimate
R$79/month
Billed monthly, cancel anytime
For high volume and documents hundreds of pages long, with plenty of OCR.
- 2,000 conversions per month
- Up to 100 MB and 2,000 pages per file
- All formats, including PowerPoint and EPUB
- Up to 50 files at a time
- 300 OCR conversions per month
- Queue for long documents, with progress
Card payment via Stripe, in Brazilian reais. No lock-in: cancel anytime from the subscription portal in your dashboard. The free plan doesn’t ask for a card.
| Feature | Free | Pro | Ultimate |
|---|---|---|---|
| Conversions per month | 30 | 500 | 2,000 |
| Input formats | PDF, DOCX, XLSX, CSV, HTML | + PPTX and EPUB | + PPTX and EPUB |
| Files per batch | 3 | 10 | 50 |
| Saved presets | |||
| Image extraction | |||
| Split by section | |||
| Max file size | 10 MB | 50 MB | 100 MB |
| Pages per file | 100 | 1,000 | 2,000 |
| OCR per month | 50 | 300 | |
| Queue for long documents | |||
| Rendered preview and source view | |||
| Interface in Portuguese, English and Spanish | |||
| File discarded after conversion | |||
| Email support |
After subscribing
What happens when you subscribe
The plan takes effect as soon as payment is confirmed, with no manual step.
Pay on Stripe
Secure checkout with a credit card, in Brazilian reais. We don’t store card data.
Limits unlocked instantly
You return to the dashboard with the new quota: more conversions, bigger files, OCR and the queue for long documents.
Everything in the portal
Change card, download invoices, switch plans or cancel from the subscription portal, anytime.
FAQ
Frequently asked questions
Why does Markdown save credits?
When you send a PDF to an LLM, each page is processed as extracted text plus a page image (1,500 to 3,000 tokens per page). The same content as plain Markdown costs 300 to 800 tokens per page and also drops repeated headers and footers that only clutter the context.
Why Markdown and not plain text?
Plain text loses what gives the document meaning: which line is a heading, what is a list item, where a table starts and ends. The model spends tokens guessing. In Markdown the structure comes along for almost nothing (a # or a |), and answers about sections and tables get more accurate.
Which formats besides PDF?
Word (.docx), Excel (.xlsx), CSV and HTML on any account, including the free one; PowerPoint (.pptx) and EPUB on the Pro and Ultimate plans. Without an account: PDF and CSV. Old .doc, .xls and .ppt files are not accepted: open them in Office and save as .docx, .xlsx or .pptx. Detection is by content, so a wrong extension is not a problem.
Can I convert several files at once?
Yes. Drop several files or a whole folder (with subfolders) and they go into a queue; each one counts as one conversion. Batch limit: 3 on the free plan, 10 on Pro and 50 on Ultimate. At the end, "Download all" builds a .zip with the same folder structure and a summary. Documents that went to the server queue keep running even if you close the tab: they show up in the dashboard for 24 hours.
Does it work with scanned PDFs?
Yes, with OCR on the Pro and Ultimate plans. Scanned PDFs are images with no text layer: the file goes into a queue and text is recognized page by page, with progress on screen. OCR uses your plan’s quota. If only some pages are scanned, we convert the rest and tell you which ones came out incomplete.
Do tables and images survive the conversion?
Tables with borders become Markdown tables, with rows and columns; borderless tables may come out as lines of text. In "Options" you can swap the table for plain text (column: value), which costs fewer tokens. Images are omitted by default, because they rarely help the model and cost vision tokens; you can leave a placeholder or, on paid plans, extract them to an images/ folder next to the .md (up to 20 per document).
What’s the size limit?
Free plan: 10 MB and 100 pages per file, 30 conversions a month and batches of 3. Pro accepts 50 MB, 1,000 pages, 500 conversions and batches of 10; Ultimate, 100 MB, 2,000 pages, 2,000 conversions and batches of 50. Documents over 80 pages go to the queue and you follow the progress.
Does it work on Mac, Linux and mobile?
Yes. PDF2MD is a website: there is nothing to install. It works on any system with a browser, including your phone. The conversion runs on our servers and the file is discarded when done.
Are my files stored?
No. Direct conversion happens in memory and the file is discarded as soon as the response is sent. Documents that go to the queue (OCR or long files) are kept only while the job runs: the file is deleted on completion and the Markdown expires in 24 hours. Server logs keep only route, status, duration and size, never the content or the file name.
How do I cancel or switch plans?
From the dashboard, with the "Manage subscription" button, which opens the Stripe portal. There you change the card, download invoices, switch plans or cancel. When you cancel, the plan stays active until the end of the period you already paid for.
Which payment methods?
Credit card, in Brazilian reais, processed by Stripe. Monthly or yearly subscriptions (yearly costs 10 monthly payments). We don’t store card data.
Do I get a receipt or invoice?
Stripe emails a receipt for every charge, and invoices are available in the subscription portal. The Brazilian service invoice (nota fiscal) is issued by the company behind PDF2MD.
Does it work with PDFs in any language?
Yes. Text extraction is language-independent. OCR recognizes Portuguese, English and Spanish. The interface is available in those three languages.
Your next PDF doesn’t have to be expensive
Drop a file and see the savings on screen. Your first conversions don’t need an account.