The best website copiers in 2026, measured
HTTrack, wget, SingleFile, monolith, SiteSucker and CopyAnySite — what each is for, with fidelity scores measured offline.
Most roundups in this category are a list of names with the vendors' own adjectives attached. This one is organised around a single question — does the copy still work once the original server is gone? — because every other quality is downstream of that.
How we measured
Eight pages chosen to cover the ways a page can be hard: a simple agency site, animation-heavy marketing, a table-layout page, a commercial WordPress theme demo, a content-heavy article, a gradient-and-JS-heavy marketing site, a page built out of SVG filters, and a docs site. Each tool copies each page; each copy is then rendered with the original host unreachable and diffed against the live page with pixelmatch, at desktop and mobile widths, median of three runs. The harness is in our repo and re-runnable with npm run bench:rivals.
Two numbers matter: the fidelity score, and how many requests the copy still aims at the original site. A copy with 200 remote requests is a bookmark.
| Tool | Avg fidelity | Remote requests left | Best on |
|---|---|---|---|
| CopyAnySite | 98.87% | 4 | 8 of 8 |
| SingleFile | 87.93% | 0 | 0 |
| wget -p -k | 87.40% | 554 | 1 |
| monolith | 83.44% | 365 | 1 |
Per page, the spread is what tells the story: on the easy pages everyone is within a point of each other (Wikipedia: 100 / 99.99 / 99.87 / 99.92), and on the hard ones the gap is 20–45 points (stripe.com: 95.99 / 65.55 / 56.86 / 49.03; the theme demo: 98.39 / 74.64 / 66.00 / 84.54).
The tools
HTTrack — the classic whole-site mirror
Free, ancient, still works. Point it at a domain, get a browsable local mirror with rewritten links. It fetches what the server sends, so on a server-rendered site it is excellent and on a React or Vue site you get a shell. GUI on Windows (WinHTTrack), CLI everywhere. Use it when you need hundreds of pages of a traditional site.
wget --mirror — the same job, in a terminal
wget --mirror --page-requisites --convert-links --adjust-extension \
--span-hosts --domains=example.com,cdn.example.com https://example.com/Installed on every Unix box, scriptable, and the flags matter enormously: without --convert-links and the right --domains, hundreds of references stay pointed at the origin — which is exactly what our measurements show (554 remote requests across eight pages). Same client-rendering limitation as HTTrack.
SingleFile — the best free single-file snapshot
A browser extension that serialises the rendered page into one HTML file with assets as data URIs. It reads the page after JavaScript, which puts it a category above Ctrl+S, and it inlines assets at any size, which is why it left zero remote requests in our runs. Its weakness is the visual layer on design-led pages — webfont adoption, pseudo-elements, masks, layered gradients — where it scored 74.64% on the theme demo and 65.55% on stripe.com. For articles and documentation it is hard to beat, and it costs nothing.
monolith — one file, from the command line
A Rust CLI that bundles a page and its assets into one HTML file. Fast, scriptable, no browser needed — and that last part is the catch: it fetches the server's HTML, so client-rendered content is missing, and in our runs it left 365 requests aimed at the origin. Best used inside pipelines on server-rendered pages.
Cyotek WebCopy — Windows, free, GUI
A friendly Windows front-end for whole-site copying with rules, filters and a link map. Same server-side fetch model, so same blind spot on JavaScript-built pages. Good for non-technical Windows users mirroring a traditional site.
SiteSucker — the macOS equivalent
Paid, native, straightforward, well-maintained. Same category, same limits.
Screenshot-to-code / AI builders
Paste a URL or image, get generated markup. Genuinely useful for inspiration and for turning a layout into components, but everything it cannot know it invents: fonts, spacing, colours, assets and the copy itself. See copying a website with AI.
CopyAnySite — ours
A Chrome extension that reads the live DOM and the computed styles of the page you have open — including pseudo-elements, hidden SVG sprites and canvas pixels — embeds the assets, and exports a standalone HTML file, a ZIP, editable WordPress/Gutenberg blocks, or a kit for a coding agent. It also crawls a whole site and ships an MCP server. It is the reason the benchmark exists, so read the numbers with that in mind — and re-run them yourself.
What it costs you: file size. Embedding everything means our stripe.com copy is 50 MB against SingleFile's 5 MB. And what it cannot do is the same thing none of these can: recreate a backend, a database, an authenticated session or DRM-protected media.
Choosing, in one line each
- Many pages of an old-school site → HTTrack or
wget --mirror. - A free, one-file snapshot of an article → SingleFile.
- Scripted bundling in a pipeline → monolith.
- Windows GUI, no terminal → Cyotek WebCopy. macOS → SiteSucker.
- A design-led page that has to look right offline, or has to land in WordPress → CopyAnySite.
- Inspiration and code scaffolding → an AI builder, then verify everything.
FAQ
What is the best free website copier?
For whole sites, wget --mirror and HTTrack, both free. For a single page as one file, SingleFile. All three are excellent within their model and all three struggle with pages that are built by JavaScript in the browser.
Is there a website copier that works online, without installing anything?
Online copiers exist, but they fetch the page from their own server, so they see neither your session nor the JavaScript-rendered result, and they are the weakest option for anything modern. A browser extension runs where the rendered page already is.
Which website copier handles JavaScript-heavy sites?
Only tools that read the browser-rendered DOM: SingleFile and CopyAnySite among those tested. Server-fetch tools (HTTrack, wget, monolith, WebCopy, SiteSucker) see the pre-JavaScript HTML.
How can I tell if a copy is any good?
Render it with the original host blocked and compare it to the live page. Then check dev tools for requests still going to the original domain. Those two checks eliminate most tools' marketing claims in about a minute.
Can any of these copy a site's backend or database?
No. Everything on this page copies the front end only. Server code, databases, APIs, authentication and personalised data are not retrievable from the browser.
Try it on the page you are looking at
CopyAnySite is a Chrome extension: open any page you can reach, press capture, and take away a standalone HTML file, an editable WordPress page, or a kit a coding agent can build from. The free key needs no card.
Download CopyAnySite