Interface EmbedTextResult

The result returned by embedText.

Example

const { embeddings, usage } = await embedText({
  provider: 'openai',
  model: 'text-embedding-3-small',
  input: ['Hello', 'World'],
});
console.log(embeddings.length); // 2
console.log(embeddings[0].length); // e.g. 1536
interface EmbedTextResult {
    embeddings: number[][];
    model: string;
    provider: string;
    usage: {
        promptTokens: number;
        totalTokens: number;
        costUSD?: number;
    };
}

Properties

embeddings model provider usage

Properties

embeddings

embeddings: number[][]

One embedding vector per input string. Each vector is a plain number[] of floats whose dimensionality depends on the model (and the optional dimensions parameter).

model

model: string

Model identifier reported by the provider (may differ from the requested model).

provider

provider: string

Provider identifier used for the run.

usage

usage: {
    promptTokens: number;
    totalTokens: number;
    costUSD?: number;
}

Token usage for the embedding request. Most embedding APIs only report prompt tokens (the input); completion tokens are typically zero.

Type declaration

  • promptTokens: number

    Number of tokens consumed by the input text(s).

  • totalTokens: number

    Sum of prompt and any other tokens (usually equal to promptTokens).

  • Optional costUSD?: number

    Cost in USD: Gemini's when it prices the call itself (embedding models with a catalog price), and an OpenAI embedding model's from OpenAI's price table. Absent when neither is known.

Settings

Member Visibility

Theme

On This Page

embeddingsmodelproviderusage