Puppeteer vs Selenium

Both are open source under Apache-2.0 and both control a real browser from code. The table uses the versions current in October 2026: Puppeteer 25.16.0 on npm and Selenium 4.51.0, released on October 9, 2026 (Selenium downloads).

The biggest technical difference is the connection. Puppeteer opens one WebSocket to the browser and sends DevTools Protocol commands over it, the same channel Chrome DevTools uses. That gives it direct access to network events, request interception, cookies, performance traces and PDF printing.

Classic Selenium sends each command as an HTTP request to a driver process (chromedriver, geckodriver, msedgedriver or safaridriver), and the driver controls the browser. That extra hop is what makes Selenium browser-neutral: the same script runs against every browser that ships a WebDriver driver, and the commands are a W3C standard. Selenium 4 adds WebDriver BiDi, a two-way protocol that brings network and log events to the standard (Selenium BiDi), and CDP access for Chromium browsers.

Puppeteer moved toward the same standard: it drives Firefox through WebDriver BiDi since v23.0.0 (Puppeteer FAQ). So the gap is narrowing, but in October 2026 Puppeteer's network features are still the more direct of the two on Chrome.

For a single Chrome scraper, Puppeteer can be faster, because there is no driver process between your code and the browser and fewer round trips per command. In practice the difference is small next to page loads: a page that takes 2 seconds to load takes 2 seconds in both. Published speed tests differ by workload, so no general figure holds; measure your own pages with the same browser, the same waits and the same proxy.

What changes run time far more than the library:

• Waits. Replace fixed sleeps with waits for a selector or a network response.
• Browser reuse. Open new pages or contexts in a running browser instead of launching a browser per URL.
• Blocking heavy resources. Images, fonts and video are most of a page's bytes. Puppeteer blocks them with request interception; in Selenium you use BiDi network interception or Chrome preferences.
• Parallelism. Selenium Grid spreads browsers over machines; with Puppeteer you run several browsers or pages yourself.

If your team does not write JavaScript, the choice is made: Puppeteer is Node.js only, and its Python port, Pyppeteer, says it is unmaintained (Pyppeteer on GitHub). Selenium's core bindings cover Java, Python, C#, Ruby and JavaScript, and its docs, answers and books exist in all of them.

Browsers matter more for testing than for scraping. Selenium covers Safari through Apple's safaridriver and Edge through msedgedriver, which Puppeteer does not. For scraping, most teams run Chromium in either tool, so browser coverage is rarely the deciding point.

Both render JavaScript, click, scroll, fill forms and read the page after its scripts have run. For a Node.js scraper on Chrome, Puppeteer is the shorter path: it installs its own browser, waits for elements with page.locator(), intercepts requests to save traffic and reads JSON responses as they arrive. Selenium is the better fit for a Python or Java team that already knows it, and for jobs that need many browsers on many machines through Grid.

The same page in both, reading the quote texts from a JavaScript-rendered practice page:

// Puppeteer (Node.js): npm i puppeteer
import puppeteer from 'puppeteer';
const browser = await puppeteer.launch();
const page = await browser.newPage();
await page.goto('https://quotes.toscrape.com/js/');
await page.waitForSelector('.quote');
const quotes = await page.$$eval('.quote .text', els => els.map(e => e.textContent));
console.log(quotes.length, quotes[0]);
await browser.close();

# Selenium (Python): pip install selenium
from selenium import webdriver
from selenium.webdriver.common.by import By
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC

options = webdriver.ChromeOptions()
options.add_argument("--headless=new")
driver = webdriver.Chrome(options=options)  # Selenium Manager finds the driver
driver.get("https://quotes.toscrape.com/js/")
WebDriverWait(driver, 10).until(EC.presence_of_element_located((By.CSS_SELECTOR, ".quote")))
quotes = [e.text for e in driver.find_elements(By.CSS_SELECTOR, ".quote .text")]
print(len(quotes), quotes[0])
driver.quit()

Before you pick either: a browser is only needed when the data appears after JavaScript runs. Many pages load their data from a JSON endpoint you can call directly, which is faster and uses a fraction of the traffic; see how to scrape hidden APIs and embedded JSON. When the HTML already holds the data, a parser such as Cheerio is enough, as compared in Cheerio vs Puppeteer.

Both send their traffic through an HTTP, HTTPS or SOCKS5 proxy; they differ in where the login goes.

• Puppeteer: pass --proxy-server=http://gate.proxyfield.io:7000 at launch, then page.authenticate({ username, password }). Details and rotation: Puppeteer proxy setup.
• Selenium with Chrome: the --proxy-server flag takes no credentials, so you either add your IP to the allowlist, load a small auth extension, or answer the login with Selenium's BiDi network auth handler (Python 4.32 and later). All three are in Selenium proxy setup.

With the ProxyField gateway, <login>-country-us-city-new_york as the username asks for an exit IP in New York. You get a new IP on every request, or a sticky session from 30 s to 36 h when a login or a cart must keep one IP.

• Node.js, Chrome, scraping or PDF and screenshot jobs: Puppeteer.
• Python, Java, C# or Ruby team: Selenium, or Playwright if you start fresh (compared in Playwright vs Selenium vs Puppeteer).
• Tests on Safari, Edge, Firefox and Chrome from one suite: Selenium with Grid.
• Network-heavy work (intercepting requests, reading API responses, blocking resources): Puppeteer.
• A working codebase in either: keep it; switch only for a missing language, browser or feature.

Step-by-step tutorials: web scraping with Puppeteer and web scraping with Selenium. For the tools themselves, see what Puppeteer is and what Selenium is.