concept

Embeddings

Embeddings are dense numeric vectors that represent the meaning of text, images, or other data, allowing similarity to be measured as vector distance.

An embedding model maps a piece of content into a vector (typically 384, 1024, or 3072 dimensions) where semantically similar inputs land close together. This unlocks semantic search (find documents about the same idea, not just the same words), clustering, classification, and RAG. Popular embedding models in 2026 include OpenAI text-embedding-3-large, Voyage AI voyage-3, Cohere embed-v3, and open weights like nomic-embed and BGE. Embeddings are paired with a vector database (Pinecone, Qdrant, Weaviate, pgvector, LanceDB) for efficient nearest-neighbor lookup.

When to use embeddings

Semantic search over text corpora.
Clustering related content (FAQs, support tickets).
Powering the retrieval step in RAG.

Common mistakes

Mixing embeddings from different models in the same index — distances are not comparable.
Skipping normalization (most use cases want L2-normalized vectors for cosine similarity).
Re-embedding the whole corpus on a prompt change — embed documents once, change the query side.

FAQ

What is embeddings?

Embeddings are dense numeric vectors that represent the meaning of text, images, or other data, allowing similarity to be measured as vector distance.

When should I use embeddings?

Semantic search over text corpora. Clustering related content (FAQs, support tickets). Powering the retrieval step in RAG.

What are the most common mistakes with embeddings?

Mixing embeddings from different models in the same index — distances are not comparable. Skipping normalization (most use cases want L2-normalized vectors for cosine similarity). Re-embedding the whole corpus on a prompt change — embed documents once, change the query side.

Retrieval-augmented generation (RAG) — Retrieval-augmented generation (RAG) injects relevant documents into the prompt at query time so the model answers from your data instead of its training memory.
Vector database — A vector database stores embeddings and performs approximate nearest-neighbor search at scale, the persistence layer behind RAG and semantic search.
Semantic search — Semantic search finds documents by meaning rather than keyword match, using embedding similarity in a vector space.

Last updated: 2026-06-01. Raw markdown: https://promtable.com/glossary/embeddings.md.