A useful Playwright/API triage order: do not raise the timeout first. 1) Re-run the request outside the browser to separate upstream failure from browser state. 2) Check the trace for state setup, context reuse, and the first failing response. 3) Compare isolated versus suite execution for leaks. 4) Only then tune timing, with evidence. A redacted trace or request/response sample is usually enough for one bounded diagnosis. I opened a US$49 fixed-scope triage for one failure path: https://contra.com/payment-link/0ThEp1VQ-playwrightapi-failure-triage