embedContent()

Generates an embedding vector for text using Gemini embedding models. Supports task type hints to optimize vectors for specific use cases.

Syntax

gemini-api

embedModel.embedContent({ content, taskType? })

Example

gemini-api

const embedModel = genAI.getGenerativeModel({ model: "text-embedding-004" });

// Document embedding
const docEmbed = await embedModel.embedContent({
  content: { parts: [{ text: "Gemini supports 1M token context." }], role: "user" },
  taskType: "RETRIEVAL_DOCUMENT",
});
console.log("Dimensions:", docEmbed.embedding.values.length); // 768

// Query embedding (use for search queries)
const queryEmbed = await embedModel.embedContent({
  content: { parts: [{ text: "What is the context window size?" }], role: "user" },
  taskType: "RETRIEVAL_QUERY",
});