---
title: "Create embeddings from input"
method: POST
path: "/semantics/vector"
tags: ["semantics-vector"]
---

# Create embeddings from input

`POST /semantics/vector`

## Request

- Base URL: `https://api.rosette.com/rest/v1`
- URL: `https://api.rosette.com/rest/v1/semantics/vector`
- Auth: API key in header `X-RosetteAPI-Key`

Provides the embedding for the input text. <br><br>Returns a single vector of floating point numbers for your input, known as a text embedding. Among other uses, a text embedding enables you to calculate the similarity between two documents or two words.

## Request body

- union
  - object
    - `content` string
    - `language` string: 3 letter code of language.
    - `options` object
      - `embeddingsMode` 'DEFAULT' | 'GEN_1' | 'GEN_2': the generation of semantic embeddings to use for all languages; defaults to second generation unless the language only supports first generation embeddings
      - `pertoken` boolean: whether to return per-token embeddings
  - object
    - `contentUri` string
    - `language` string: 3 letter code of language.
    - `options` object
      - `embeddingsMode` 'DEFAULT' | 'GEN_1' | 'GEN_2': the generation of semantic embeddings to use for all languages; defaults to second generation unless the language only supports first generation embeddings
      - `pertoken` boolean: whether to return per-token embeddings

## Response `200`

OK

- SemanticVectorsResponse
  - `documentEmbedding` number[]: document embeddings
  - `tokenEmbeddings` number[]: per-token embeddings
  - `tokens` string[]: extracted tokens

---

[API](https://skmtc.dev/babelstreet/apis/rosette-text-analytics.md) · [All operations](https://skmtc.dev/babelstreet/apis/rosette-text-analytics/llms.txt) · [OpenAPI document](https://skmtc.dev/babelstreet/apis/rosette-text-analytics/revisions/7c1d1497a5dd?raw)