GlossaryRetrieval
Embeddings
Dense numeric vectors that represent text in a high-dimensional space where similar texts are close.
Embeddings are vectors (typically 768-3,072 dimensions) produced by a model trained to place semantically related text close together. They power semantic search: convert the query and every document to embeddings, then compute cosine similarity. Modern embedding models include OpenAI's text-embedding-3-large, Voyage-3, Cohere v3, and BGE-M3.