Google DeepMind has released EmbeddingGemma 2, an open multimodal embedding model designed for efficient on-device applications. This model unifies text, images, video, and audio into a single 768-dimensional vector space, with a total of 740 million parameters. It offers multilingual capabilities, improved code understanding, and flexible modality loading, allowing developers to use only the necessary components. EmbeddingGemma 2 also supports Matryoshka Representation Learning for reduced storage costs and features an 8K token context window. AI
IMPACT Enables efficient on-device multimodal AI applications, potentially lowering latency and cost for RAG and search.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
Read on Hugging Face Trending Models →
- Atomic Chat
- Docker Model Runner
- GitHub
- Google DeepMind
- Hugging Face
- Jan
- Lemonade
- llama.cpp
- LM Studio
- Ollama
- OpenAI
- unsloth/embeddinggemma-2-GGUF
- Winget
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →