Build with the API

Create embeddings

POST /v1/embeddings turns text into vectors you can search, compare or cluster. It takes an embedding model, which is one whose capabilities include embeddings in the model list.

curl https://sven.swedishembedded.com/v1/embeddings \
  -H "Authorization: Bearer $SVEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "MODEL_ID", "input": ["Hello", "World"], "encoding_format": "float"}'

Parameters#

model is required.
input is one text, or a list of up to 2048 texts. Each text must not be empty.
encoding_format may only be float, which is also what you get when you leave it out. OpenAI's clients ask for a different encoding unless you say otherwise, so pass float explicitly as shown.
dimensions is not supported and is refused with a 400.

The response#

A list with one embedding per input, in order, as an array of numbers, each with its index, and the usage, in input tokens:

json

{
  "object": "list",
  "model": "MODEL_ID",
  "data": [{ "object": "embedding", "index": 0, "embedding": [0.0123, -0.0456, 0.0789] }],
  "usage": { "prompt_tokens": 2, "total_tokens": 2 }
}

What it costs#

You are charged per input token, never more than the model's price. See How calls are charged.