Your app can now search text, images, and audio together — for free

If you have ever wished your app could search through documents, photos, and voice recordings all at once, a new free tool from Google makes that possible today. Google DeepMind released EmbeddingGemma 2 on October 6 — a small, open AI model that understands text, images, video, and audio together as one system.

Here is how it works. The model reads any content — a sentence, a photo, a sound clip — and turns it into a list of numbers that captures its meaning. Two things that mean the same thing get similar numbers. So when a user searches for 'dog barking outdoors,' the model can match that query to a relevant photo, an audio clip, and a text description — all in a single search step, without needing three separate systems.

The practical side is just as good. The full model fits in about 567 MB of RAM, small enough to run on a modern phone. It is released under Apache 2.0, which means it is completely free to use in commercial products. It supports over 100 languages and is already available on Hugging Face and Kaggle.

The old way to build multimodal search required separate AI models for each media type, separate indexes, and custom code to combine results. That added cost, complexity, and API bills. Now, one small open model does the job. For any team building search, recommendations, or content matching, the biggest obstacle just got removed — and there is no reason to wait to try it.