# Google shows a community classifier running on EmbeddingGemma 2 in real time

> Original title: People are finding really creative ways to use EmbeddingGemma 2 beyond search.

- Source: Google Gemma
- Published: 2026-11-04T17:44:29.000Z
- Bad Influence: https://badinfluence.news/items/phtggtyas481ugcisl1vid5cj
- Original: https://x.com/googlegemma/status/2118036675540689044

## Summary

Google says a community project trains a lightweight classifier on top of EmbeddingGemma 2 in real time, instead of running a full multimodal LLM. The post says the model can learn new categories on the fly and classify inputs locally in about 20ms for text and 100ms for images.

## Post

People are finding really creative ways to use EmbeddingGemma 2 beyond search.

Instead of running a full multimodal LLM, this community project trains a lightweight classifier on top of EmbeddingGemma 2 in real time.

Because the model is compact and natively multimodal, it can learn new categories on the fly and classify inputs locally in ~20ms for text and ~100ms for images.

[Video](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Famplify_video_thumb%2F2118036266939977728%2Fimg%2FH2AQN3ce0XBnB5Du.jpg&mode=thumb&exp=1794009600&sig=12277e62721b7b18)