2026-10-06 # Google DeepMind releases EmbeddingGemma 2 multimodal embedding model On October 6, 2026, Google DeepMind launched EmbeddingGemma 2, a 740-million-parameter model built on the Gemma 4 architecture under an Apache 2.0 license, which maps text, images, audio and video into one embedding space for on-device use, per deepmind.google. The company says the model leads sub-1B multimodal embedders on benchmarks including MTEB Code and MAEB, a claim based on its own figures, not independently verified. ## What else the source says - Prior EmbeddingGemma model surpassed 20 million downloads - EmbeddingGemma 2 has 740 million parameters - Model is built on the Gemma 4 architecture - Released under an Apache 2.0 commercial license - Uses technology shared with Gemini Embedding models Source. [deepmind.google](https://deepmind.google/blog/embeddinggemma-2-an-open-lightweight-multimodal-embedding-model/) ## TRACE - First seen Oct 6, 2026, 1:29 PM PT - Published Oct 7, 2026, 12:33 PM PT, 23.1 h after first seen - Source deepmind.google (Press) [SOTO home](https://signalsoftomorrow.com/)