POST /v1/embeddings — Swytcho Text Embeddings API

The /v1/embeddings endpoint converts one or more text strings into high-dimensional numeric vectors. These embeddings capture semantic meaning, so texts with similar meaning produce vectors that are close together in the vector space. Use embeddings to power semantic search, document clustering, recommendation systems, and retrieval-augmented generation (RAG) pipelines.

Endpoint

POST https://api.swytcho.com/v1/embeddings

Request Parameters

string

required

The ID of the embedding model to use. Embedding models are separate from chat and completion models — call GET /v1/models and filter for models with object: "embedding" to see your options.

string | array

required

The text(s) to embed. Pass a single string to embed one piece of text, or an array of strings to embed multiple texts in a single request. Each string is embedded independently; the order of the output data array matches the order of your input.

string

The format of the returned embedding vectors. Use "float" (default) to receive a standard JSON array of floating-point numbers, or "base64" to receive a base64-encoded binary blob, which reduces response payload size for large batches.

Request Examples

curl https://api.swytcho.com/v1/embeddings \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "swytcho-embed-v1",
    "input": "The quick brown fox jumps over the lazy dog.",
    "encoding_format": "float"
  }'

import os
from swytcho import Swytcho

client = Swytcho(api_key=os.environ["SWYTCHO_API_KEY"])

response = client.embeddings.create(
    model="swytcho-embed-v1",
    input="The quick brown fox jumps over the lazy dog.",
    encoding_format="float",
)

vector = response.data[0].embedding
print(f"Embedding dimensions: {len(vector)}")

import Swytcho from "swytcho";

const client = new Swytcho({ apiKey: process.env.SWYTCHO_API_KEY });

const response = await client.embeddings.create({
  model: "swytcho-embed-v1",
  input: "The quick brown fox jumps over the lazy dog.",
  encoding_format: "float",
});

const vector = response.data[0].embedding;
console.log(`Embedding dimensions: ${vector.length}`);

Response Fields

string

Always "list".

string

The embedding model ID that was used to generate the vectors.

array

An array of embedding objects, one per input string, in the same order as your input array.

Show Embedding object fields

integer

Zero-based index corresponding to the position of this embedding in the input array.

string

Always "embedding".

array | string

The embedding vector. When encoding_format is "float" (default), this is a JSON array of numbers. When encoding_format is "base64", this is a base64-encoded string.

object

Token usage statistics for the request.

Show Usage object fields

integer

Number of tokens across all input strings.

integer

Total tokens consumed (same as prompt_tokens for embeddings — there are no completion tokens).

Example Response

{
  "object": "list",
  "model": "swytcho-embed-v1",
  "data": [
    {
      "index": 0,
      "object": "embedding",
      "embedding": [0.0023064255, -0.009327292, 0.01589467, -0.004121838, 0.0082586454]
    }
  ],
  "usage": {
    "prompt_tokens": 9,
    "total_tokens": 9
  }
}

The number of dimensions in the returned vector depends on the embedding model you choose. Check the model details via GET /v1/models/{model_id} for the dimensions field before designing your vector store schema — storing vectors with the wrong dimension will cause index errors. Changing models after populating a vector store requires re-embedding all existing records.

Overview

Endpoints

POST /v1/embeddings — Swytcho Text Embeddings API

Endpoint

Request Parameters

Request Examples

Response Fields

Example Response