Browser-based vs upload-based file tools
Short answer: most online file tools send your document to a server, and this is measurable rather than a matter of opinion. On 2026-08-11, 12 of 20 online PDF compression tools were observed transmitting a complete 3.00 MB test file to a named host. Server-side processing is not wrong — it handles larger files, works the same on any device, and does some conversions better. It just means the privacy claim rests on the operator’s policy instead of on what the software is able to do.
This page is published by Filevane, which is one of the tools being compared. That is a reason to check it rather than to trust it, so every claim about another product below comes from a dated transport-layer measurement whose raw data is downloadable, and the method is published so you can re-run it on any of them — this site included.
Measured: tools that uploaded the test file
Each row records bytes actually transmitted after the file was selected, and where they went. A measurement describes the tool on the date given; architectures change.
| Tool | Processing | Bytes sent | Destination |
|---|---|---|---|
| Filevane (control) | Client-side | 0 | Nothing transmitted |
| Smallpdf | Server-side | 3,156,480 | smallpdf-production-files…r2.cloudflarestorage.com |
| iLovePDF | Server-side | 2,107,638 | api112.ilovepdf.com |
| PDF2Go | Server-side | 3,156,480 | www35.pdf2go.com |
| Sejda | Server-side | 3,156,480 | www.sejda.com |
| Soda PDF | Server-side | 3,156,480 | api-gw.sodapdf.com |
| PDF Candy | Server-side | 3,156,480 | s33.api.pdfcandy.com |
| PDF24 Tools | Server-side | 3,156,480 | filetools23.pdf24.org |
| FreePDFConvert | Server-side | 3,156,480 | www.freepdfconvert.com |
| DocFly | Server-side | 3,156,480 | create-pdf.s3-eu-west-1.amazonaws.com |
| TinyWow | Server-side | 3,156,480 | tinywow.com |
| CleverPDF | Server-side | 3,156,480 | www.cleverpdf.com |
| PDFSimpli | Server-side | 3,156,480 | prodlegalsimplistorage.blob.core.windows.net |
Not independently verified
No valid measurement was obtained for these 7 tools. That is a fact about the measurement, not about the tool. None of them should be read as uploading or as not uploading; they are listed because omitting them would make the sample look cleaner than it was.
| Tool | Why no result |
|---|---|
| Adobe Acrobat online | The page would not load from our test network across three attempts (HTTP/2 protocol error). The hostname also resolved to an address outside Adobe's published ranges, which suggests DNS interference on our connection rather than anything about the service. |
| AvePDF | DNS resolution failed repeatedly from our test network; the page never loaded. |
| LightPDF | The page loaded and accepted the file, but we could not confirm the tool ever started, so the absence of an upload proves nothing. |
| Xodo | The page loaded and accepted the file, but we could not confirm the tool ever started. Outbound traffic in the window was third-party telemetry, not the document. |
| PDFgear | We could not confirm the tool ever started. |
| OnlineConvertFree | No file input could be located automatically; this one needs a manual pass. |
| DocuPub | The file was accepted but we could not confirm the tool ever started. This row is the reason the 'did it actually run' check exists: without it, this tool would have been recorded as processing locally. |
When a server is the better choice
Uploading is a trade, not a defect. These are the cases where it is the right one.
- Very large files
- Browser processing is bounded by the memory the tab is allowed. A server is bounded by the machine, which is far larger. For a 500 MB scanned archive, a server-side tool is often the only one that finishes.
- Low-powered devices
- Client-side work runs on the visitor's hardware. On an old phone, a WebAssembly OCR pass that takes six seconds on a laptop can take a minute or more; a server does it at the same speed for everyone.
- Heavier conversions
- Faithful PDF-to-Word conversion, high-quality OCR over many languages, and video transcoding at scale are all better on a server today. Browser implementations exist and are usable, but the output is not always equivalent.
- Cross-device workflows
- An account and server storage let you start on a phone and finish on a desktop. A tool that keeps everything in the tab cannot offer that, because there is nothing to sync.
When the browser is the better choice
- The file has nowhere to leak from
- A file that is never transmitted cannot be exposed by a breach at the processor, retained past a stated deletion window, or handed over in response to a request. This is a structural property, not a policy.
- No upload wait
- On a typical home connection, sending a 50 MB PDF takes longer than most local operations do. Removing the transfer removes the slowest step.
- It works offline
- Once the page has loaded, a fully client-side tool keeps working with the network disconnected — which is also the simplest way for a visitor to prove where the processing happens.
- The claim is checkable
- A server-side tool's privacy policy can only be believed. A client-side tool's behaviour can be observed in the browser's own Network tab in about thirty seconds.
How to decide
- Confidential document? Use a tool whose behaviour you can observe. Whether it is this one does not matter — the check takes thirty seconds and works on any site.
- Very large file, or an old phone? A server-side tool will usually finish when a browser-based one will not.
- Faithful PDF-to-Word, or heavy multi-language OCR? Server-side output is still generally better. Try the browser version first and compare the result on your own document rather than on anyone’s claim.
- Routine task, ordinary file? Either works. The browser-based one will usually be faster, because there is no upload to wait for.
Filevane’s own position, stated plainly
Filevane processes files in the browser, and 405 of its 406 tools contain no network API anywhere in their source. The exception is the Currency Converter, which sends the two currency codes you selected in order to fetch a rate; the amount you typed is not transmitted. Three OCR tools download a language model the first time you use a given language — data comes in, none goes out. The site also runs Google Analytics, which records page views and has no access to files opened in a tool.
Filevane does not do faithful PDF-to-Word conversion as well as the best server-side tools, and very large files are limited by the memory your browser gives the tab. Those are real costs of the architecture and there is no version of this page where they are not.
Sources and corrections
Every measurement on this page comes from the PDF Upload Audit of 2026-08-11, published with its method and raw data under CC BY 4.0. Operators of any tool named here have a standing right of reply; if a result is wrong, or has become wrong, it will be corrected and the correction dated.