Working notes on embedding caching, cost, and invalidation from Skelf Research.https://embedcache.skelfresearch.com/en-ushttps://embedcache.skelfresearch.com/blog/embedding-spend-is-a-tax-on-stable-inputs/https://embedcache.skelfresearch.com/blog/embedding-spend-is-a-tax-on-stable-inputs/Most of the budget you burn on hosted embedding APIs pays for vectors you've already computed. A cache turns that line item into a one-time cost.Tue, 12 May 2026 00:00:00 GMThttps://embedcache.skelfresearch.com/blog/content-hash-vs-lru/https://embedcache.skelfresearch.com/blog/content-hash-vs-lru/Two cache strategies for embeddings — one keyed on content, one keyed on access recency. They optimize for different things, and only one of them is correct for stable corpora.Wed, 08 Apr 2026 00:00:00 GMThttps://embedcache.skelfresearch.com/blog/stale-while-revalidate-for-embeddings/https://embedcache.skelfresearch.com/blog/stale-while-revalidate-for-embeddings/SWR works well for HTTP, less obviously for embeddings. A pragmatic walk through when it helps, when it bites, and what to do about model migrations.Tue, 03 Mar 2026 00:00:00 GMT