Glossary · Data collection

Headless browser

A headless browser is a real web browser engine running without a visible window and controlled by code instead of a person; it loads pages, executes JavaScript and builds the DOM like a normal browser, which makes it the tool for pages whose content only appears after scripts run.

Reviewed 30 Oct 2026

How a headless browser is driven

Chrome runs headless with the --headless flag. Since Chrome 112 the headless mode is the full browser without a displayed window, and since Chrome 132 the older implementation ships separately as chrome-headless-shell.

Code controls it over a remote protocol — the Chrome DevTools Protocol, or WebDriver, which the W3C specifies as a remote control interface for user agents. Most teams use a library on top: Playwright, Puppeteer or Selenium.

Headless browser vs HTTP client

	HTTP client	Headless browser
Runs JavaScript	No	Yes
Downloads	The documents you request	The page plus every script, image, font and call it triggers
CPU and memory	Minimal	A browser process per worker
Use when	Data is in the HTML or a JSON endpoint	Data appears only after rendering

The download row is the expensive one. The HTTP Archive's Web Almanac measured a median page weight of 2,652 KB on desktop and 2,311 KB on mobile in October 2024, which a browser pulls in full unless you block resources. That is why per-GB pricing punishes rendering pages you could have fetched as HTML.

Using proxies with a headless browser

Set the proxy at launch or per browser context, with username and password as separate fields where the library supports it.
Give each context its own sticky session, so a page and all its subresources leave from one exit.
Block images, media and fonts you do not need; the reduce proxy bandwidth guide shows how.

FAQ

Questions

Something not covered? Write to contact@arcproxies.io.

What is a headless browser used for?

Automated testing, generating screenshots and PDFs, and collecting data from pages that render their content with JavaScript. It is also the usual execution environment for AI agents that browse the web.

Do headless browsers use more bandwidth than HTTP requests?

Yes, usually by a wide margin, because the browser downloads every script, stylesheet, image and font the page references. Blocking unneeded resource types and reusing a cache cut most of the difference.

Can websites tell a headless browser from a person?

Often, yes. Automated browsers differ from human sessions in many ways besides the network. A proxy changes only the exit address; it does not make automation invisible, and it is not meant to.

How do I use an authenticated proxy in Playwright?

Pass the proxy at launch or when creating a context, with server, username and password as separate fields. With a rotating pool, put the session handle in the username so each context keeps its own exit.

Sources & review

Last reviewed 30 October 2026 by the ARC Proxies team. Prices are read from the live catalog at build time.

[1]Chrome for Developers: Chrome Headless mode
[2]W3C: WebDriver
[3]HTTP Archive: Web Almanac 2024: Page Weight

In practice

Where headless browser matters

IntegrationPlaywrightSet up an authenticated proxy in Playwright for Python and Node: launch and per-context proxies, sticky sessions, geo-targeting and handling proxy errors.IntegrationPuppeteerUse an authenticated proxy in Puppeteer: --proxy-server plus page.authenticate, per-context proxies for sticky sessions, geo-targeting and 407 handling.Integrationbrowser-useGive a browser-use agent its own exit IP: ProxySettings on the Browser, one sticky session per task, country targeting, a pre-flight check and retries.Use caseResidential proxies for AI agentsBring your own residential IPs to browser-use, Playwright, Crawl4AI, n8n and any MCP tool — one sticky session per task, billed per GB.GuideHow to reduce proxy bandwidthCut proxy bandwidth by fetching HTML or JSON instead of rendering, blocking images and fonts, using compression, conditional requests and local caching.