Camoufox vs Patchright: Firefox Fork or Patched Chromium? (2026)

Last updated: 09 Nov 2026

Camoufox vs Patchright: Firefox Fork or Patched Chromium? (2026)

Short answer: start with Patchright. It keeps the Playwright API, runs on Chrome, and the switch is mostly an import line. Move to Camoufox only when Patchright is set up the way its README recommends and a target still blocks you, and the target is worth running a heavier browser for.

The two tools patch different layers of different browsers. Camoufox changes a whole Firefox build so the values a detection script reads are set inside the engine. Patchright changes the Playwright driver so Chrome stops announcing that it is automated. Camoufox does more work per browser and costs more per page. Patchright does less, and less is usually enough until it isn't.

Facts below are from the Camoufox and Patchright repositories, checked Nov 2026.

Choose Before You Read

Your situation Use
Node.js stack, Cloudflare blocking you Patchright
Patchright set up as its README recommends, still blocked Camoufox
Tens of thousands of pages a day Patchright first: do the compute maths before Camoufox
A handful of high-value pages on a hard target Camoufox
Target behaves differently in Firefox Patchright
Blocked by behavioural scoring, not fingerprinting Neither fixes this
You want the data, not a stealth stack See the last section

Two Different Patch Layers

Both tools try to stop a site from noticing automation without injecting a JavaScript stealth plugin into the page. They do it in different places.

Patchright keeps Playwright's API and patches the Playwright driver, not the Chromium binary. Its README calls avoiding the Runtime.enable CDP call its biggest patch; it also changes launch flags (adds --disable-blink-features=AutomationControlled, drops --enable-automation and a few others) and disables Console.enable, so console output does not work. It supports Chromium-based browsers only and recommends real Chrome over Chromium. If you have not installed it yet, the install commands and its language and browser coverage are in What Is Patchright?

Camoufox is a Firefox fork. Its README says fingerprint data "is intercepted at the C++ implementation level", so values such as screen, WebGL, fonts and navigator properties are set inside the browser rather than overwritten by a script after the page loads. There is no patched JavaScript property for a detection script to spot, because the change happened before the page ran.

The difference in approach explains the difference in cost. Patchright removes the signs that say this is Playwright. Camoufox ships a modified browser that presents a consistent made-up identity.

Camoufox vs Patchright: Side by Side

Every row below comes from the project's own repository (checked Nov 2026). Neither project publishes independent detection or speed figures, and we do not quote any.

Patchright Camoufox
Browser Chromium-based only; real Chrome recommended Firefox fork
API Playwright, drop-in Playwright-compatible, Firefox
What is patched The Playwright driver: Runtime.enable, launch flags, Console.enable The Firefox engine itself, at C++ level
Browser binary Unmodified Chrome or Chromium Its own patched Firefox build to download
Memory A normal Chrome per context About 200MB, per its README
Known caveats Console output disabled README warns it "has gone down in performance due to the base Firefox version" and may not suit stable production use
Behavioural scoring Not addressed Not addressed
Licence Apache-2.0 MPL-2.0
Best fit Volume on mid-tier targets Low volume on hard targets

The Compute Bill Decides It

Run your own numbers before choosing, because they usually decide for you.

Neither project publishes a pages-per-hour figure you can plan on, and any number from someone else's test site will not match your targets. So time it yourself: run 50 real pages from your target through each tool and note the average seconds per page, including challenge pages and retries. Then:

  • Pages per worker per day = 86,400 ÷ your average seconds per page
  • Workers needed = pages per day ÷ pages per worker per day, plus headroom for failures
  • Memory = workers × memory per browser context

The inputs differ between the two tools in a predictable direction. Camoufox runs a full patched Firefox in every context, its README puts that at about 200MB, and the same README warns about performance on the current base Firefox. Patchright runs ordinary Chrome with a patched driver, so it costs what Chrome costs. A slower, heavier browser multiplies through every line of the calculation above.

So the question is not "which is more stealthy". It is how many pages, how often, and what is a miss worth. A weekly pull of a few hundred competitor prices from one hard target can carry Camoufox's overhead. A nightly refresh of a large catalogue usually cannot, and is better split: Patchright for most of it, Camoufox only where Patchright fails.

Firefox Is Not a Free Swap

Camoufox is a Firefox fork, and that matters beyond the stealth question.

Target sites behave differently across engines. Layout, JavaScript quirks and rendering differences mean selectors validated in Chromium are not guaranteed in Firefox, and a site that quietly serves a different code path to Firefox can need separate extraction logic. If you already have a working Chromium scraper, adopting Camoufox means rewriting the parts that touch the DOM, not just swapping the launcher.

Maintenance matters too. Camoufox's README says the project had a year-long maintenance gap and is still under development. Any fork has to keep pace with Firefox's own releases, so there can be a lag between a new detection method and a patched build. Patchright carries the same kind of risk on Playwright releases, on a smaller codebase.

Neither One Fixes Behaviour

Both tools deal with technical fingerprinting. Modern bot protection also scores behaviour, and neither one touches it.

Mouse-movement paths, scroll speed, time before the first interaction, typing rhythm, the timing of requests across a session: a perfectly fingerprinted browser that loads a page, reads the DOM and leaves still looks like a script. On the strictest configurations, behaviour is what catches both tools.

Neither gives you proxies either. A clean fingerprint arriving from a datacentre IP that has already been flagged gets refused before any of the patching matters.

The wider comparison, including playwright-extra, the Python playwright-stealth line and nodriver, is in Playwright stealth tools compared. If you are choosing between vanilla Playwright and a patched build in the first place, start with Patchright vs Playwright.

The Cost Neither One Removes

Both are libraries, and libraries go stale. Anti-bot vendors ship detection updates, each update needs a matching patch, and the gap between the two is downtime on whichever data source you depend on. That cycle is the recurring cost, and it does not show up in any install command.

If what you need is competitor prices, stock levels and product data in your own systems rather than a stealth stack to keep alive, this comparison may be the wrong one. ScrapeWise (which publishes this blog) runs the browsers, the proxy rotation and the challenge handling and delivers structured rows as a table, CSV/Excel, API or MCP. It does not get through every site; some block automated visits or need a login. There is no plan and no compute units: you top up a balance and pay per delivered page, EUR 0.15 per 1,000 for plain requests, EUR 0.75 with rendering, EUR 1.50 through residential IPs and EUR 3.75 for both. Schedules run up to once a day, and you can also run a job by hand.

If the end goal is comparing your prices with competitors, you can upload your own product list and match competitor rows to it in the same account: on a shared EAN or ID first, then on title, brand and size, then on product pictures, with unsure pairs left for you to review.

You can also call 35 ready-made scraping API endpoints directly, or see how this feeds competitor price tracking and brand protection.

So Which Should You Pick?

  • Patchright: the default. Drop-in Playwright API on real Chrome, no separate browser build to maintain. Start here and find out whether you need more.
  • Camoufox: Patchright is set up as its README recommends and still blocked, the target is a hard one, and the volume is low enough that a heavier Firefox per worker is affordable on your own numbers.
  • Both: Patchright for most of the catalogue, Camoufox kept for the few targets that refuse it. Common, and cheaper than running Camoufox across everything.
  • Neither: you want delivered rows on a schedule and would rather not maintain a stealth stack at all.

Before you size a browser fleet, run the URLs that block you through ScrapeWise: the first 5 requests are free, no card needed. If you would rather have the whole job done for you, Managed Data and pricing are here.

Paste a URL your current tool cannot reach

Run the pages your current tool misses and see what comes back. Pay per page, no per-SKU pricing.

Not ready to sign up? See 40 rows from a real Amazon search →

5 free requests no credit card · pay per page, no monthly plan · book a 15-min call →

FAQ

Frequently asked questions

Camoufox vs Patchright in 2026: what each one patches, what it costs to run, whether Firefox is a free swap, what neither one fixes, and which fits your volume

They patch different things. Camoufox is a Firefox fork that changes fingerprint values at the C++ level inside the browser. Patchright patches the Playwright driver, not the browser: it avoids the Runtime.enable CDP call, changes launch flags and disables Console.enable, and it runs on Chromium-based browsers only, with real Chrome recommended. Camoufox does more and costs more per browser; its README puts memory at about 200MB and warns that performance has dropped on the current base Firefox. Start with Patchright and move to Camoufox only when Patchright, set up as its README recommends, still gets blocked.

Neither project publishes a throughput figure you can plan on, and numbers from someone else's test page will not match your targets. Time 50 of your own pages and take the average seconds per page, including retries. Pages per worker per day is 86,400 divided by that number; divide your daily page count by it for the workers you need, and multiply workers by memory per context (about 200MB for Camoufox, per its README). That calculation usually shows why Camoufox suits low-volume, high-value targets better than a nightly catalogue refresh.

Not cleanly. Camoufox is a Firefox fork, and target sites behave differently across engines. Layout, JavaScript quirks and rendering differences mean selectors validated in Chromium are not guaranteed in Firefox, and a site that serves a different code path to Firefox can need separate extraction logic. Adopting Camoufox on an existing Chromium scraper means rewriting the parts that touch the DOM, not just swapping the launcher.

Both deal with technical fingerprinting; neither deals with behaviour. Modern bot protection scores mouse-movement paths, scroll speed, time before the first interaction, typing rhythm and the timing of requests. A perfectly fingerprinted browser that loads a page, reads the DOM and leaves still looks like a script. Neither tool gives you proxies either, and a flagged datacentre IP is refused before any patching matters.

Yes, and it is usually cheaper than running Camoufox across everything. Patchright handles most of the catalogue on ordinary Chrome, and Camoufox is kept for the few targets that refuse it, typically the ones where a block means losing a data source rather than a page.

Both tools are libraries, and libraries go stale: anti-bot vendors ship detection updates, each needs a matching patch, and the gap is downtime on whichever source you depend on. If what you need is competitor prices, stock levels and product data in your own systems, a scraping service removes that work. ScrapeWise, which publishes this blog, runs the browsers, proxies and challenge handling and delivers structured rows, from EUR 0.15 per 1,000 pages for plain requests up to EUR 3.75 with rendering and residential IPs. It can also match competitor rows to your own product list. Schedules run up to once a day, some sites still block automated visits or need a login, and every new account gets 5 free requests with no card.