Skip to main content

Gemini

To use Gemini embeddings, you need to import GeminiEmbedding from llamaindex.

import { GeminiEmbedding, Settings } from "llamaindex";

// Update Embed Model
Settings.embedModel = new GeminiEmbedding();

const document = new Document({ text: essay, id_: "essay" });

const index = await VectorStoreIndex.fromDocuments([document]);

const queryEngine = index.asQueryEngine();

const query = "What is the meaning of life?";

const results = await queryEngine.query({
query,
});

Per default, GeminiEmbedding is using the gemini-pro model. You can change the model by passing the model parameter to the constructor. For example:

import { GEMINI_MODEL, GeminiEmbedding } from "llamaindex";

Settings.embedModel = new GeminiEmbedding({
model: GEMINI_MODEL.GEMINI_PRO_LATEST,
});