Cypress runs your test code inside the browser; Playwright drives the browser from outside over a WebSocket. That single difference explains most of what teams feel in CI — worker memory footprint, how many workers fit on a runner, and which flows (multi-tab, cross-origin) are natively supported. This is an editorial explainer of documented architecture, not a lab benchmark.
Jump to section
Execution model
In-browser vs out-of-process
Documented architecture difference
Isolation unit
Browser instance vs context
Per official docs
Pricing
Both open source
Cloud orchestration is the paid part