Quantable

Extract tables from research papers

No more retyping. Go from PDF to data, accurately and easily. Extract, review, then export to CSV or paste into Excel, Sheets, or a notebook. Works on the toughest PDFs: borderless LaTeX tables, nested and multi-level headers, skewed scans of old journals, scientific notation, footnote markers.

JOIN THE WAITLIST

How it works

fig. 1
getquantable.com / okafor-2024-thermal-drift.pdf · p. 4demo · no PDF uploaded
Thermal drift in low-noise amplifier arrays under passive shielding
R. Okafor · M. Lindqvist · Dept. of Instrumentation
Fig. 2 — Drift response, 0–40 °C.
Table 3. Measured drift by condition
ConditionnµV/°CSDp
Baseline240.42 ± 0.080.08—
Shielded240.31 ± 0.060.06.04
Cooled220.18 ± 0.050.05<.001
Shield + Cool†220.11 ± 0.040.04<.001
† Combined treatment; n reduced by sensor dropout.
RESULT — editable before exportwaiting for a selection
ConditionnµV/°CSDp
Baseline240.42 ± 0.080.08—
Shielded240.31 ± 0.060.06.04
Cooled220.18 ± 0.050.05<.001
Shield + Cool†220.11 ± 0.040.04<.001
  1. STEP 1
    Open the paper
    Drop in the PDF, whether it's an arXiv preprint or a scan of a 1980s journal.
  2. STEP 2
    Select the table
    Draw a box to select exactly the data you're interested in
  3. STEP 3
    Review the cells
    An editable grid with ± values, footnote markers, and results exactly as printed.
  4. STEP 4
    Export the data
    Download CSV or copy as TSV straight into Excel, Sheets, R, or a notebook.

Specifications

accuracy
Optimized for the tables that break generic PDF tools: cells land in the right row and column, numbers aren't merged or dropped, and values are reproduced exactly. Benchmark results on real papers coming soon.
messy input
Skewed scans, low-resolution photocopies, tables that run into the margin, watermarks over the data. Extraction works as well as it does on a clean preprint.
layouts
Two-column pages, tables spanning the gutter, captions and footnotes nearby. Draw the box and only the table comes out.
LaTeX tables
Borderless tables (the norm in arXiv and journal PDFs) extract as cleanly as fully ruled ones.
structure
Nested and multi-level headers, merged cells, row groups, and sub-tables come out as a proper grid.
values
± errors, footnote markers, significance stars, units, and Greek symbols stay intact in their cells.
scans
Scanned older papers are handled with class-leading OCR.
review
Every extraction opens as an editable preview: correct a cell, split a ± column, then export.
output
CSV download · copy as TSV into Excel, Sheets, or pandas · .xlsx coming soon · LaTeX tabular coming soon

Frequently asked questions

Q. How do I extract a table from a research paper?
Open the paper's PDF in Quantable, draw a box around the table, check the editable result, and download it as CSV or copy it into your spreadsheet or notebook.
Q. Does it work with arXiv / LaTeX PDFs?
Yes. Borderless booktabs-style tables, the standard in LaTeX papers, extract with their multi-level column headers intact.
Q. Can it handle scanned older papers?
Yes. Scans and native PDFs go through the same extraction, so a skewed, photocopied 1970s journal article works like a fresh preprint. The editable preview is there for the odd cell that needs a fix.
Q. What happens to ± values and footnote markers?
They stay in the cell exactly as printed: “0.31 ± 0.06” and “†” don't get split or dropped. If you'd rather have the error in its own column, you can split it in the editable preview before exporting.
Q. Is my PDF stored?
No. Your PDF never leaves your browser. Only a cropped image of the table you select is sent to our extraction service, and we don't keep it.
Q. Can I export to Excel or LaTeX?
Copy-as-TSV pastes cleanly into Excel and Sheets today, and CSV loads into R, pandas, or anything else. Native .xlsx export and LaTeX tabular output are coming soon.
Q. How accurate is it?
It's built and evaluated on real scientific tables: nested headers, merged cells, ± columns, rough scans. Cells land in the right place and values come through as printed, and the editable preview lets you confirm before you export. Benchmark figures are coming soon.
Q. Does it handle nested or multi-level headers?
Yes. Column groups, sub-headers, merged cells, and row-group labels are preserved as spans in the extracted grid rather than collapsed into a single header row.
Q. Can I just extract a table from any PDF?
Yes. Quantable is built and tested on scientific papers, but the same flow works on any PDF table, such as a report, a thesis appendix, or a supplementary data file.
Q. When can I use it?
Join the waitlist. Early access invitations go out in small batches, researchers first.