OOmnistu

Blog / Engineering Playbooks

Engineering Playbooks

Playwright vs Cypress: how to choose your end-to-end test runner

Both are current and open source: Playwright drives Chromium, Firefox and WebKit builds from outside the app, while Cypress runs inside the browser in the same run loop as your app, with tests in JavaScript or TypeScript. How the two test runners differ on browsers, debugging, CI and hosted services - from both projects' own documentation, read 21 November 2026.

OOmnistu
Reviewed 2026-11-21 · 4 min read

Two names dominate end-to-end web testing in late 2026, and both are shipping: Playwright 1.65.1 (13 November 2026) and Cypress 16.3.0 (10 November 2026), per their GitHub release pages. Both are open source - Apache-2.0 and MIT respectively, per their repository licenses. Choosing between them is not about which is "better"; it is about two genuinely different models of how a test talks to your app. This article compares them from both projects' own documentation and release posts, all read 21 November 2026, and names no default winner.

The core difference: inside the app or outside it

Cypress's own documentation states its architecture in one sentence: "Cypress executes in the same run loop as your application," and it does not use Selenium or WebDriver. Your test code runs inside the browser alongside the app, which is what enables its signature features: time-travel snapshots of every command, debugging with the browser's own DevTools, automatic waiting for commands and assertions, and network stubbing without touching your server. Playwright takes the outside position: its install docs describe a framework that drives Chromium, WebKit and Firefox builds - on Windows, Linux and macOS, headless or headed - with native mobile emulation for Chrome on Android and Mobile Safari. The practical split shows up at the edges: Cypress's own trade-offs page notes that its commands run inside a browser, so a test is bound to a single superdomain (cross-origin navigation goes through its cy.origin command) and new-tab flows need a plugin, while Playwright's external model handles multiple pages, origins and browser engines natively.

Browsers and languages, scoped exactly

Playwright supports Chromium, WebKit and Firefox per its documentation - the WebKit line means Playwright's own WebKit builds plus emulation for Mobile Safari, which is engine-level coverage, not testing on a real iPhone or the shipping Safari binary. On languages: Playwright's automation library ships bindings for JavaScript/TypeScript, Python, Java and .NET (the Python and other docs are library-plus-test-plugin guides), while the Playwright Test runner itself runs on Node.js. Cypress's detailed browser guide lists Chrome-family browsers (including Edge and Chrome for Testing) and Firefox, with WebKit marked experimental: enabling it takes an experimentalWebKitSupport: true config flag and installing the playwright-webkit package, and Cypress presents it as the practical way to validate Safari behavior from Windows, Linux or CI without a Mac. Its FAQ is explicit that "you write your tests in JavaScript or TypeScript," whatever language the app under test uses. Neither list is a guarantee that your app behaves identically across engines - it is a statement of where the runner can execute, per each project's docs as read today.

Authoring and debugging

Cypress's pitch, per its documentation, is the tight local loop: time travel through command snapshots, readable errors, automatic waiting (no manual sleeps), spies/stubs/clocks, network interception, and Cypress Studio with Studio AI to record interactions and suggest assertions. Playwright's equivalent kit from its docs: the code generator (codegen) that records actions into test code, the Trace Viewer that replays a recorded trace with DOM snapshots and network logs, a UI mode for watch-style development, auto-waiting locators and web-first assertions, and an optional GitHub Actions workflow offered at install time. Both projects have invested heavily here, so judge this section by trying both on one real flow rather than by feature lists.

CI, scale and hosted services - current status

Playwright's test runner builds in parallelization and sharding, so scaling on your own CI needs no extra product. On hosted services, get the current facts: Microsoft's retirement notice says Microsoft Playwright Testing was retired on 8 March 2026, and that Playwright users should create a Playwright Workspace in Azure App Testing, which is now generally available - so there is a first-party-adjacent hosted path, under a new name, if you want one. Cypress's hosted story is Cypress Cloud: per its docs, Test Replay re-runs recorded CI runs for zero-configuration debugging, Smart Orchestration handles parallelization and spec prioritization, and Flake Detection tracks unreliable tests; Cypress also sells premium Accessibility and UI Coverage products. The Cypress app itself is the open-source runner; the Cloud is a separate hosted service with its own plans, and its docs describe a free trial - check current terms on their site rather than assuming either way.

Which one to pick

Tie it to your situation. Team writes JavaScript/TypeScript, tests center on Chrome-family browsers, and a visual, in-browser debugging loop is the priority: Cypress's model is built for exactly that, with Cypress Cloud as the optional hosted layer. You need coverage across Chromium, Firefox and WebKit engines, tests in Python/Java/.NET as well as TS, built-in parallelization without a hosted dependency, or heavy multi-tab/multi-origin flows: Playwright's model fits. Already running a suite on either one: neither project's docs, checked today, document an official converter, so moving means porting tests by hand - but that can happen incrementally, one critical flow at a time alongside the existing runner, rather than as a single rewrite. And if you are genuinely undecided, write your single most critical user flow in both and compare the daily loop: that afternoon tells you more than any comparison article.

Sources, all read 21 November 2026: Playwright documentation, Installation (framework scope, Chromium/WebKit/Firefox, platforms, mobile emulation); Playwright documentation, Browsers (browser builds and emulation); Cypress documentation, Why Cypress (same-run-loop architecture, no Selenium/WebDriver, time travel, automatic waiting, Studio AI; Cypress Cloud, Accessibility and UI Coverage products; page updated 19 November 2026); Cypress documentation, Launching Browsers (Chrome-family and Firefox support, experimental WebKit via config flag and the playwright-webkit package); Cypress documentation, Trade-offs (single superdomain per test, cy.origin for cross-origin, plugin for new tabs, simulated events); Cypress documentation, FAQ (tests are written in JavaScript or TypeScript); Cypress Cloud documentation (documented free trial); Microsoft Learn, Playwright Testing retirement banner (retirement 8 March 2026; Playwright Workspaces in Azure App Testing now generally available; page updated 18 December 2025); GitHub, playwright releases (v1.65.1 of 13 November 2026, Apache-2.0) and GitHub, cypress releases (v16.3.0 of 10 November 2026, MIT).

Frequently asked questions

What are the current versions of Playwright and Cypress?

As read on 21 November 2026 from their GitHub release pages: Playwright v1.65.1 (13 November 2026) and Cypress v16.3.0 (10 November 2026). Both are open source - Apache-2.0 and MIT respectively per their repository licenses.

Does Playwright test on real Safari or real iPhones?

Playwright tests on WebKit - Safari's browser engine - using Playwright's own WebKit builds, and offers emulation for Mobile Safari, per its documentation. That is engine-level coverage, not the shipping Safari binary on a Mac and not a physical iPhone.

Can Cypress test Safari and Firefox?

Per Cypress's current browser guide: Chrome-family browsers (including Edge and Chrome for Testing) and Firefox are supported, and WebKit is experimental - it requires an experimentalWebKitSupport config flag and installing the playwright-webkit package. Cypress frames WebKit testing as the practical way to validate Safari behavior from Windows, Linux or CI without a Mac; it is still engine-level coverage, not the shipping Safari binary.

Is Cypress free to use?

The Cypress app is open source under the MIT license per its repository. Cypress Cloud is a separate hosted service with its own plans - its documentation describes a free trial, and current terms live on Cypress's site. Cypress also sells premium Accessibility and UI Coverage products.

Is Playwright free, and is there a hosted option?

Playwright is open source under Apache-2.0 with the runner's parallelization built in. For hosted runs, Microsoft Playwright Testing was retired on 8 March 2026; Microsoft's notice directs users to Playwright Workspaces in Azure App Testing, now generally available.

Can I convert an existing Cypress suite to Playwright (or back)?

Neither project's documentation, checked 21 November 2026, documents an official converter. Moving means porting tests by hand, which teams typically do incrementally - new or migrated flows on the new runner while the old one keeps covering the rest.