EmbeddingGemma
An open-source embedding model supports text, code and media search locally and offline on consumer CPUs or GPUs under Apache 2.0.
EmbeddingGemma is a local AI embedding model for developers adding search and retrieval to applications. It represents text, images, audio and video in a shared space so apps can find related content across media types.
- Search across media: Search text, source code, images, audio and video by meaning, including matches across different media types.
- Retrieval for chatbots: Supply relevant content to multimodal retrieval-augmented generation (RAG) applications, with Gemma 4 E2B as a companion model for generating responses.
- Zero-shot classification: Use the model for classification tasks without training it on examples of each target category.
- Adjustable embeddings: Reduce the size of stored embeddings to save vector database space, with some loss of quality.
- Multilingual text and code: Process text and source code across over 100 languages, with input support up to 8,192 tokens.
The model runs locally and offline on consumer CPUs or GPUs. Embedding generation can stay on your hardware without sending inputs to a cloud service. It produces embeddings rather than conversational replies.
EmbeddingGemma is open source, and its weights use the Apache 2.0 license. You can fine-tune the model and deploy it in your own applications.
Logotyper
