Flags SDK and Vercel Flags now evaluate multiple feature flags in bulk up to 10x faster, with gains that scale as flag count increases.
The speedup comes from two targeted fixes: reduced microtask queue overhead and fewer promise allocations. The change requires switching from Promise.all([flagA(), flagB()]) to the new await evaluate([flagA, flagB]) call. The evaluate function also accepts a named-key object for structured flag sets, and precompute() picks up the same improvements automatically. The mechanism behind the microtask reduction is worth reading in the full documentation if you run high-flag-count environments.
Upgrade the flags and @flags-sdk/vercel packages to the latest versions to get the improvement. No architectural changes required, just a one-line call site update.
[READ ORIGINAL →]