EmbeddingGemma is a 308M parameter multilingual text embedding model based on Gemma 3. It is optimized for use in everyday devices, such as phones, laptops, and tablets. The model produces numerical representations of text to be used for downstream tasks like information retrieval, semantic similarity search, classification, and clustering.
EmbeddingGemma includes the following key features:
- Multilingual support: Wide linguistic data understanding, trained in over 100 languages.
- Flexible output dimensions: Customize your output dimensions from 768 to 128 for speed and storage tradeoffs using Matryoshka Representation Learning (MRL).
- 2K token context: Substantial input context for processing text data and documents directly on your hardware.
- Storage efficient: Run it on less than 200MB of RAM with quantization
- Low latency: Generative embeddings in less than 22ms on EdgeTPU for fast and fluid applications.
- Offline and secure: Generate embeddings of documents directly on your hardware, works without internet connection to keep sensitive data secure.
Get it on Hugging Face Get it on Kaggle Access it on Vertex
As with other Gemma models, EmbeddingGemma is provided with open weights and licensed for responsible commercial use, allowing you to fine tune and deploy it in your own projects and applications.