Best practices for cache TTL on tiny GET endpoints?

Looking for community wisdom on edge cache TTL settings for tiny pure-function GET endpoints. I'm running an academic calculator at gpacalc.app where state never changes but exam-season traffic spikes 50x. Anyone tried Mlytics edge functions for this read-heavy static-ish workload? Currently using 24h s-maxage but wondering if shorter TTL with stale-while-revalidate is actually safer for spike scenarios.