“Your files are processed in your browser” is claimed by most online file tools and measured by almost none of them. Filevane publishes the measurements: 20 PDF tools tested at the transport layer on 2026-08-11, of which 12 were observed transmitting the entire 3.00 MB test document to a server — together with the method, the raw data under CC BY 4.0, and instructions for running it against any site, including this one.
Only work that reports a measurement: a stated method, a date, a scope, and numbers someone else could reproduce or contradict. Explainers live in the guides, not here. A research index padded with articles is not a research index.
Of 20 online PDF compression tools tested on 2026-08-11, 12 were measured transmitting the complete 3.00 MB test document to a server, 1 processed it without sending it, and 7 produced no valid measurement.
6 methods for establishing whether a web page transmits a file you give it, each stating what it can and cannot establish, from a 30-second check in the browser's Network tab to the byte-counting harness used in the audit.
The audit measurements are published as a CSV and a JSON file under CC BY 4.0, alongside a README stating the method, the three verdict rules and the meaning of every column.
Every row is dated because a tool can change its architecture at any time, and a measurement from August is a claim about August. If a result here is wrong, or has become wrong, the page will be corrected and the correction dated rather than quietly edited. Operators of any tool named in the audit can get in touch — there is a standing right of reply, and the audit page sets out how it works.