Embedding model

An embedding model turns text, images or other data into a list of numbers that captures its meaning, so that similar items end up close together.

  • Models and architecture
  • Updated

Embeddings power semantic search, recommendations and retrieval-augmented generation. You embed your documents once, embed each question, and return the documents whose numbers lie closest. An embedding model does not write text; its output is a vector.

Embedding models are small compared with chat models. Qwen3-Embedding ranges from 0.6B to 8B parameters, and EmbeddingGemma 2 has fewer than one billion, so both run on ordinary laptops.

Related terms

  • Retrieval-augmented generation (RAG) Retrieval-augmented generation lets a model answer from your own documents by retrieving the relevant passages and adding them to the prompt.
  • Multimodal model A multimodal model can handle more than one type of data, such as text, images, audio or video, as input, as output, or both.

Keep reading

All terms