embeddings API
An API that converts text into a vector (embedding), useful for similarity search, clustering, and semantic retrieval.
Think of it like
A fingerprint—instead of storing the whole face, store a short code that captures its essence.
Example
Call the API with 'The cat sat on the mat' and get back a 1536-dim vector. Call it again with 'A feline rested on fabric.' Vectors are similar; high cosine similarity means semantic match.
How it actually works
Embeddings come from the model's internal representations. Different models have different dimension sizes (OpenAI 1536, newer ones smaller). Embeddings are how you enable semantic search (RAG's backbone). Once you have embeddings, store them in a vector DB, search by distance.
For product teams
Core for RAG; enables semantic search without ever running a full LLM.
For engineers
Fast (usually <100ms), small output, critical for retrieval pipelines.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome