Playwright vs Selenium WebDriver

Free and open source

Free and open source

Free and open source. Microsoft also sells a hosted parallel-runner service; the library itself has no paid tier.

Free and open source. The real cost is grid infrastructure or a device-cloud subscription.

Everything, self-hosted

Everything, self-hosted

Category standard

Microsoft · since 2020

Category standard

Software Freedom Conservancy · since 2004

• Auto-waiting and web-first assertions remove most of the sleep() flake that plagues older tools.
• The trace viewer — DOM snapshots, network, console and screenshots per step — is the best failure triage in the category.
• True WebKit support, so Safari bugs are caught on Linux CI without a Mac.

• The widest language and browser support of anything in the category, including legacy browser versions.
• A W3C standard, so it will outlive any single vendor. Skills transfer between employers.
• Every device cloud speaks Selenium, which keeps you portable between vendors.

• No native mobile app support. Mobile means an emulated viewport, not a real device.
• The API surface is large; teams without a conventions document end up with four ways to write the same locator.
• Component testing is still experimental compared to the E2E runner.

• No built-in waiting strategy, reporting, parallelism or assertions — expect to assemble and maintain a framework around it.
• Explicit waits are your problem, and getting them wrong is the number one source of flake in real Selenium suites.
• Out-of-process communication makes it slower than Playwright or Cypress.

Teams that write their own tests and want the lowest-flake open-source option available.

Enterprises with polyglot teams, legacy browser requirements, or an existing grid investment.