Embedding storage size calculator

Estimate how much memory and disk an embedding table needs: vectors × dimensions × bytes per number, plus an index overhead factor for approximate vector DB size.

Number of documents (vectors)

Average tokens per document

Informational for planning; storage math uses one vector per document.

Embedding dimensions

Precision

Vector DB / index overhead multiplier

Typical planning range ~1.2–2.0 depending on HNSW/IVF and metadata.