Skip to main content

Voyage AI

Maven Dependency​

<dependency>
    <groupId>dev.langchain4j</groupId>
    <artifactId>langchain4j-voyage-ai</artifactId>
    <version>1.21.0-beta31</version>
</dependency>

APIs​

  • VoyageAiEmbeddingModel

Capabilities​

  • Multimodal (voyage-multimodal-3, voyage-multimodal-3.5 — auto-detected from the model name): embeds text and images into a shared vector space; interleaved text + image is fused into a single embedding. Provide image inputs as ImageContent (URL or base64) in an EmbeddingRequest.
  • Per-call parameters: input_type (EmbeddingInputType.QUERY / DOCUMENT), and Voyage AI's own truncation and encoding_format via VoyageAiEmbeddingRequestParameters. A value set on the builder is the default for every call, and a value set on the request overrides it for that call. truncation(false) makes a call fail instead of shortening an input that exceeds the model's context. encoding_format ("base64") only asks for a smaller response; the embeddings you get back are the same, and only the text models accept it, so a multimodal model rejects it.
  • Listeners: configure via VoyageAiEmbeddingModel.builder().listeners(...).

See Embedding Model for the request/response API and multimodal usage.

Examples​