Technique
Embeddings Cache
Local store keyed by content hash so repeated inputs aren't re-embedded — cuts cost.
Definition
Embedding APIs charge per token, so caching results by content hash avoids paying twice. The cache typically lives in Redis or a key-value store keyed by SHA-256 of the input plus model name. Standard pattern in RAG ingestion pipelines.
Common use cases
- Ingestion pipelines
- Cost reduction
- Idempotency