# Google Gemma releases EmbeddingGemma 2, a 740M-parameter multimodal embedding model

> Original title: EmbeddingGemma 2 maps text, images, video, audio, and code, running 100% locally on your device.

- Source: Google Gemma
- Published: 2026-10-29T16:54:38.000Z
- Bad Influence: https://badinfluence.news/items/zjn9xr6rz525iwc5zyo596f6a
- Original: https://x.com/googlegemma/status/2115849803636871174

## Summary

Google Gemma announces EmbeddingGemma 2, which maps text, images, video, audio, and code into embeddings. The model has 740M parameters and runs 100% locally on the device, including in the browser via WebGPU and Transformers.js. The post names multimodal search, cross-modal classification, and clustering as example uses.

## Why it matters

The post gives the model's size and browser-based runtime path, which show how multimodal embeddings can run locally without a server.

## Post

EmbeddingGemma 2 maps text, images, video, audio, and code, running 100% locally on your device.

At just 740M parameters, you can build powerful multimodal apps running entirely in your browser via WebGPU & Transformers.js. From multimodal search to cross-modal classification and clustering, local embeddings unlock endless possibilities.

[Video](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Famplify_video_thumb%2F2115848539708510208%2Fimg%2FOiqe_--AxpA527dm.jpg&mode=thumb&exp=1793491200&sig=bbc215aaae899780)