Skip to main content
The default Voyage AI provider instance. This is a pre-configured instance created with createVoyage() that uses the VOYAGE_API_KEY environment variable.

​
Import

import { voyage } from 'voyage-ai-provider';

​
Provider methods

The Voyage provider supports multiple methods for creating different model types:

​
textEmbeddingModel()

Creates a text embedding model for generating embeddings from text inputs.
​
modelId
VoyageEmbeddingModelId
required
The ID of the text embedding model to use. Supported models include:
  • voyage-3.5 - Latest flagship model
  • voyage-3.5-lite - Lighter version of 3.5
  • voyage-3-large - Large context model
  • voyage-3 - Standard model
  • voyage-3-lite - Lightweight model
  • voyage-code-3 - Code embeddings
  • voyage-finance-2 - Finance domain
  • voyage-multilingual-2 - Multilingual support
  • voyage-law-2 - Legal domain
  • voyage-code-2 - Code embeddings (older)
  • Or any custom model ID string
Returns: EmbeddingModelV3 - A text embedding model instance

​
Example

import { voyage } from 'voyage-ai-provider';
import { embed, embedMany } from 'ai';

const model = voyage.textEmbeddingModel('voyage-3-lite');

const { embedding } = await embed({
  model,
  value: 'The quick brown fox jumps over the lazy dog',
});

const { embeddings } = await embedMany({
  model,
  values: [
    'Artificial intelligence is transforming the world',
    'Machine learning enables computers to learn',
  ],
});

​
imageEmbeddingModel()

Creates an image embedding model for generating embeddings from image inputs.
​
modelId
VoyageMultimodalEmbeddingModelId
required
The ID of the multimodal embedding model to use. Supported models:
  • voyage-multimodal-3 - Multimodal model for images
  • Or any custom model ID string
Returns: EmbeddingModelV3 - An image embedding model instance

​
Example

import { voyage } from 'voyage-ai-provider';
import { embedMany } from 'ai';

const model = voyage.imageEmbeddingModel('voyage-multimodal-3');

// Single image per embedding
const { embeddings } = await embedMany({
  model,
  values: [
    { image: 'https://example.com/image1.jpg' },
    { image: 'https://example.com/image2.jpg' },
  ],
});

// Multiple images in one embedding
const result = await embedMany({
  model,
  values: [
    {
      image: [
        'https://example.com/image1.jpg',
        'https://example.com/image2.jpg',
      ],
    },
  ],
});

​
multimodalEmbeddingModel()

Creates a multimodal embedding model for generating embeddings from combined text and image inputs.
​
modelId
VoyageMultimodalEmbeddingModelId
required
The ID of the multimodal embedding model to use. Supported models:
  • voyage-multimodal-3 - Multimodal model for text and images
  • Or any custom model ID string
Returns: EmbeddingModelV3 - A multimodal embedding model instance

​
Example

import { voyage } from 'voyage-ai-provider';
import { embed, embedMany } from 'ai';

const model = voyage.multimodalEmbeddingModel('voyage-multimodal-3');

// Text and image combination
const { embedding } = await embed({
  model,
  value: {
    text: ['A beautiful sunset over the beach'],
    image: ['https://example.com/beach.jpg'],
  },
});

// Multiple text-image combinations
const { embeddings } = await embedMany({
  model,
  values: [
    {
      text: ['Golden sunset over ocean waves'],
      image: ['https://example.com/beach1.jpg'],
    },
    {
      text: ['Vibrant sunset over tropical beach'],
      image: ['https://example.com/beach2.jpg'],
    },
  ],
});

// Text only with multimodal model
const textOnly = await embedMany({
  model,
  values: [
    'Customer service inquiry',
    'Technical support request',
  ],
});

// Grouped text inputs
const grouped = await embedMany({
  model,
  values: [
    [
      'Premium Wireless Bluetooth Headphones',
      'Experience superior sound quality',
      'Battery life: 30 hours',
    ],
  ],
});

​
reranking()

Creates a reranking model for reordering documents based on relevance to a query.
​
modelId
VoyageRerankingModelId
required
The ID of the reranking model to use. Supported models:
  • rerank-2.5 - Latest reranking model
  • rerank-2.5-lite - Lighter version of 2.5
  • rerank-2 - Standard reranking model
  • rerank-lite-2 - Lightweight version of 2
  • rerank-1 - First generation model
  • rerank-lite-1 - Lightweight version of 1
  • Or any custom model ID string
Returns: RerankingModelV3 - A reranking model instance

​
Example

import { voyage } from 'voyage-ai-provider';
import { rerank } from 'ai';

const model = voyage.reranking('rerank-2.5');

const { ranking } = await rerank({
  model,
  query: 'talk about rain',
  documents: [
    'sunny day at the beach',
    'rainy day in the city',
  ],
  topN: 1,
  providerOptions: {
    voyage: {
      returnDocuments: true,
      truncation: true,
    },
  },
});

​
rerankingModel()

Alias for reranking(). Creates a reranking model with identical functionality.
​
modelId
VoyageRerankingModelId
required
The ID of the reranking model to use.
Returns: RerankingModelV3 - A reranking model instance

​
Calling as a function

The provider instance can be called directly as a function, which is equivalent to calling textEmbeddingModel(). 
import { voyage } from 'voyage-ai-provider';
import { embed } from 'ai';

// These are equivalent
const model1 = voyage('voyage-3-lite');
const model2 = voyage.textEmbeddingModel('voyage-3-lite');

const { embedding } = await embed({
  model: model1,
  value: 'Hello world',
});

Build docs developers (and LLMs) love

Get started for freeTalk to us