Vector DB Sizing Calculator

Calculate vector database storage, HNSW overhead, and embedding costs for RAG systems.

Number of Documents / Files

Avg Chunks per Document: 15

150100

Embedding Dimensions

Quantization

Replication Factor

HNSW Index Overhead

Daily Update Rate: 5%

Total Chunks

150K

Raw Vectors

585.9 MB

Metadata

28.6 MB

HNSW Overhead

175.8 MB

Per Replica

790.3 MB

Total (1x replication)

790.3 MB

Full Index Cost

$15

One-time embedding

Daily Re-index Cost

$0.758K chunks/day

Monthly Re-index Cost

$23

Estimates as of March 2026. Embedding cost assumes API pricing at ~$0.0001 per chunk (text-embedding-3-large). Self-hosted embedding is significantly cheaper at scale. Metadata estimate is ~200 bytes per chunk (file path, language, symbols, timestamps). HNSW overhead is typically 20-40% of vector storage.