runtime-cache

You can cache API responses in Vercel serverless functions by storing computed results in an ephemeral per-region key-value store. This reduces latency by serving subsequent requests directly from the regional cache.

Tag-based cache invalidation for serverless workloads is a mechanism that allows granular cache purging by associating keys with tags. It enables you to selectively invalidate cached responses across per-region deployments without clearing the entire store.

Yes, runtime cache is accessible from Vercel Functions, Routing Middleware, and Builds. It integrates with @vercel/functions to provide framework-agnostic caching for session data and computed results across your deployment workflow.

Per-region isolation provides separate ephemeral key-value stores for each serverless region. This ensures cached data is localized, reducing cross-region latency while maintaining independent TTL-based eviction and LRU management for each location.

Ephemeral cache for serverless session data is limited by its temporary nature, relying on LRU eviction and TTL expiration to manage storage capacity. Data is not persisted permanently and may be evicted unexpectedly during high-traffic periods.