Skip to main content
The VoyageMultimodalEmbeddingOptions type defines the configuration options you can pass when generating embeddings with multimodal embedding models that support both text and images.

​
Properties

​
inputType
'query' | 'document'
Type of the input.Default: queryWhen inputType is specified as query or document, Voyage automatically prepends a prompt to your inputs before vectorizing them, creating vectors more tailored for retrieval/search tasks.For retrieval/search purposes where a query is used to search through documents, we recommend specifying whether your inputs are queries or documents. Since inputs can be multimodal, “queries” and “documents” can be text, images, or an interleaving of both modalities.The following prompts are prepended:
  • For query: “Represent the query for retrieving supporting documents: ”
  • For document: “Represent the document for retrieval: ”
​
outputEncoding
'base64'
The data type for the resulting output embeddings.Default: null (embeddings are represented as a list of floating-point numbers)
  • If null (not specified): The embeddings are represented as a list of floating-point numbers.
  • If base64: The embeddings are represented as a Base64-encoded NumPy array of single-precision floats.
See Voyage AI FAQ on quantization
​
truncation
boolean
Whether to truncate the input texts to fit within the context length.Default: true

​
Usage

​
Text-only embedding

import { voyage } from '@voyageai/ai-provider';
import { embed } from 'ai';

const { embedding } = await embed({
  model: voyage.multimodalEmbeddingModel('voyage-multimodal-3'),
  value: 'A beautiful sunset over the ocean',
  experimental_providerOptions: {
    voyage: {
      inputType: 'query',
    },
  },
});

​
Image embedding

import { voyage } from '@voyageai/ai-provider';
import { embed } from 'ai';
import { readFileSync } from 'fs';

const imageBuffer = readFileSync('./sunset.jpg');

const { embedding } = await embed({
  model: voyage.imageEmbeddingModel('voyage-multimodal-3'),
  value: {
    image: imageBuffer,
  },
  experimental_providerOptions: {
    voyage: {
      inputType: 'document',
    },
  },
});

​
Multimodal embedding with text and image

import { voyage } from '@voyageai/ai-provider';
import { embed } from 'ai';
import { readFileSync } from 'fs';

const imageBuffer = readFileSync('./product.jpg');

const { embedding } = await embed({
  model: voyage.multimodalEmbeddingModel('voyage-multimodal-3'),
  value: [
    { type: 'text', text: 'Product specifications:' },
    { type: 'image', image: imageBuffer },
  ],
  experimental_providerOptions: {
    voyage: {
      inputType: 'document',
      truncation: true,
    },
  },
});

​
Using base64 encoding

import { voyage } from '@voyageai/ai-provider';
import { embed } from 'ai';
import { readFileSync } from 'fs';

const imageBuffer = readFileSync('./image.png');

const { embedding } = await embed({
  model: voyage.multimodalEmbeddingModel('voyage-multimodal-3'),
  value: {
    image: imageBuffer,
  },
  experimental_providerOptions: {
    voyage: {
      inputType: 'document',
      outputEncoding: 'base64',
    },
  },
});

// embedding is now a base64-encoded string

​
Batch multimodal embeddings

import { voyage } from '@voyageai/ai-provider';
import { embedMany } from 'ai';
import { readFileSync } from 'fs';

const image1 = readFileSync('./image1.jpg');
const image2 = readFileSync('./image2.jpg');

const { embeddings } = await embedMany({
  model: voyage.multimodalEmbeddingModel('voyage-multimodal-3'),
  values: [
    { image: image1 },
    { image: image2 },
    'Text-only query',
  ],
  experimental_providerOptions: {
    voyage: {
      inputType: 'query',
      truncation: true,
    },
  },
});

Build docs developers (and LLMs) love

Get started for freeTalk to us