Class EmbeddingInput

java.lang.Object
dev.langchain4j.model.embedding.request.EmbeddingInput

@Experimental public class EmbeddingInput extends Object
A single input to embed: one or more ordered Content parts that are fused into one Embedding.

A plain text input is simply a single TextContent. Multimodal models can accept an interleaving of text and image/video/audio parts (e.g. Cohere Embed v4, Voyage multimodal) — the whole interleaved list produces a single vector.

An EmbeddingRequest carries a list of EmbeddingInputs (the batch dimension), each of which carries a list of Content parts (the interleaving dimension).

Since:
1.18.0