Interface EmbeddingCreateParams

interface EmbeddingCreateParams {
    input:
        | string
        | string[]
        | number[]
        | number[][];
    model:
        | "text-embedding-ada-002"
        | string & {}
        | "text-embedding-3-small"
        | "text-embedding-3-large";
    dimensions?: number;
    encoding_format?: "base64" | "float";
    user?: string;
}

Properties

input model dimensions? encoding_format? user?

Properties

input

input:
    | string
    | string[]
    | number[]
    | number[][]

Input text to embed, encoded as a string or array of tokens. To embed multiple inputs in a single request, pass an array of strings or array of token arrays. The input must not exceed the max input tokens for the model (8192 tokens for text-embedding-ada-002), cannot be an empty string, and any array must be 2048 dimensions or less. Example Python code for counting tokens.

model

model:
    | "text-embedding-ada-002"
    | string & {}
    | "text-embedding-3-small"
    | "text-embedding-3-large"

ID of the model to use. You can use the List models API to see all of your available models, or see our Model overview for descriptions of them.

Optionaldimensions

dimensions?: number

The number of dimensions the resulting output embeddings should have. Only supported in text-embedding-3 and later models.

Optionalencoding_format

encoding_format?: "base64" | "float"

The format to return the embeddings in. Can be either float or base64.

Optionaluser

user?: string

A unique identifier representing your end-user, which can help OpenAI to monitor and detect abuse. Learn more.

Settings

Member Visibility

On This Page

Properties