PocketClawvol. 1 · 2026

Embeddings

Numerical vector representations of text, images, or audio that capture semantic similarity.

Embeddings are typically 384-3072 dimensions and are produced by dedicated embedding models (OpenAI text-embedding-3-large, BGE, jina-embeddings-v2, etc.). Self-hosted setups in 2026 commonly use BGE-small running locally (no cloud API call needed for indexing).

Related terms

Vector databaseRAG (Retrieval-Augmented Generation)

Found a definition that's wrong, dated or could be sharper? Email us — we update with attribution unless you'd rather we didn't.