Reference · API
OpenAI Embeddings API
Turn text into vectors for semantic search and RAG.
Turn text into vectors for semantic search and RAG.
#Endpoint
POST https://api.openai.com/v1/embeddings#Request
| Field | Purpose |
|-------|---------|
| `model` | e.g. `text-embedding-3-small` |
| `input` | String or array of strings |
#Python example
from openai import OpenAI
client = OpenAI()
emb = client.embeddings.create(
model="text-embedding-3-small",
input="Annual refund policy for pro plans",
)
vector = emb.data[0].embedding
print(len(vector), "dimensions")#Usage in RAG
1. Embed all document chunks once → store vectors + text
2. Embed user question at query time
3. Cosine similarity → top-k chunks → LLM prompt
#Tips
- Batch inputs to save requests
- Keep the **same model** for index and queries
- Store raw text alongside vectors for prompt packing
#Try it yourself
Change the input text and re-run — this executes real Python against a small local stand-in for the OpenAI client, so you can see the response shape (a vector plus its length) without an API key.
Try it yourself — runs in your browser. Simulated response, not a live API call.