Embedding
An embedding is a list of numbers (a vector) that represents a piece of text, an image or other data, produced so that items with similar meaning end up with vectors close together.
Also called: vector embedding, embeddings, embedding model
By Divya Sharma · Checked Nov 15, 2026 · How we review
Computers cannot compare meaning directly, but they can compare numbers quickly. An embedding turns "How do I reset my password?" and "I forgot my login" into two vectors that sit near each other, even though they share almost no words. That is what makes semantic search, clustering and recommendations work.
Embedding models
An embedding model is the neural network that produces these vectors. You send it text (or an image, for multimodal models) and it returns a fixed-length vector. Examples include text-embedding-3-large and Voyage 4 Large. The MTEB leaderboard compares them, and the embedding models guide helps choose one.
How they are used
In retrieval-augmented generation, every chunk of your documents is embedded once and stored in a vector database. At query time the question is embedded with the same model, and the closest stored vectors are returned. Closeness is usually measured with cosine similarity or a dot product.
Things to know
Vectors from different models are not comparable, so switching models means re-embedding everything. The length of the vector is the embedding dimension, which drives storage cost. An embedding captures broad meaning well but can miss exact identifiers such as product codes, which is one reason teams add keyword search in a hybrid search setup.
Related terms
- Vector database
- A vector database stores embeddings alongside their source records and finds the stored vectors closest to a query vector, usually with an approximate index so search stays fast across millions of items.
- Embedding dimension
- Embedding dimension is the number of values in each vector an embedding model outputs. More dimensions can hold more detail, but every extra value adds storage, memory and search cost per item.
- Cosine similarity
- Cosine similarity measures how alike two vectors are by the angle between them. It ranges from -1 to 1, ignores vector length, and is the most common way to compare embeddings in search.
- Vector search
- Vector search finds items by comparing their embeddings to the embedding of a query and returning the closest ones, so results match on meaning rather than on shared keywords.
- MTEB
- MTEB, the Massive Text Embedding Benchmark, is a public benchmark and leaderboard that scores text embedding models across many task types, such as retrieval, clustering, classification and reranking, over many datasets and languages.