EmbeddingGemma

Name: EmbeddingGemma
Author: Google DeepMind

by Google DeepMind

Embeddings

EmbeddingGemma is Google DeepMind's compact text-embedding model, built from Gemma 3 and sized to run comfortably on a laptop or a small GPU. It was trained on over 100 languages and produces 768-dimension vectors that can be truncated to 512, 256, or 128 dimensions via Matryoshka representation learning, letting you trade a little accuracy for a smaller, faster index. A practical default for local RAG, semantic search, and clustering.

Publisher: Google DeepMind
API model name: embeddinggemma:300m
Context window: 2K tokens
Sizes: 300M
Licence: Gemma

Run EmbeddingGemma

Install it on a Pendra worker, then call it through the OpenAI-compatible API with a pdr_sk_ key.

from pendra import Pendra

client = Pendra(api_key="pdr_sk_...")

response = client.embeddings.create(
    model="embeddinggemma:300m",
    input="The quick brown fox",
)

print(response.data[0].embedding)

import { Pendra } from "pendra";

const client = new Pendra({ apiKey: "pdr_sk_..." });

const response = await client.embeddings.create({
  model: "embeddinggemma:300m",
  input: "The quick brown fox",
});

console.log(response.data[0].embedding);

Run EmbeddingGemma on your own UK infrastructure

Deploy a worker, install EmbeddingGemma, and start serving it through one sovereign API endpoint.

Get started free Read the quickstart

Related models

Gemma 4

Google DeepMind