Google DeepMind Launches EmbeddingGemma 2 for On-Device Multimodal Embeddings
Released under the Apache 2.0 license, the 740M-parameter model unifies text, code, images, audio, and video into a single shared vector space.
Google DeepMind has introduced EmbeddingGemma 2, an open-weights embedding model engineered specifically for efficient on-device multimodal representation. Built upon the Gemma 4 architecture, the model maps text, code, images, video, and audio into a single 768-dimensional embedding space, facilitating cross-modal retrieval, classification, and retrieval-augmented generation (RAG).
The architecture totals 740 million parameters, integrating a 270M text backbone with modular vision (170M) and audio (300M) encoders. With quantization, the complete multimodal model operates within roughly 567MB of active RAM on mobile hardware, while text-only workloads require as little as 191MB, enabling responsive local inference without an internet connection.
EmbeddingGemma 2 also features an expanded 8,192-token context window capable of handling extended audio clips and multiple images simultaneously. Distributed under the permissive Apache 2.0 license, the model provides developers with a production-ready foundation for private, low-latency search and AI agents on consumer devices.
Evidências das fontes 6
A verificação de fontes não garante exatidão permanente. Considere a fonte original, a data e as correções posteriores.
Google AI for Developers Google AI for Developers # EmbeddingGemma 2 model card Hugging Face | GitHub | Launch Blog | Documentation | License: Apache 2.0 | Authors: Google DeepMind EmbeddingGemma 2 is an open multimodal embedding model built by Google DeepMind which maps text (incl. code), images, video, and audio inputs—and combinations thereof—into a single, unified 768-dimensional vector space. The model has 740M total parameters, combining a 270M parameter text model with modular vision (170M) and audio (300M) encoders. Designed to run on consumer hardware such as mobile devices and laptops, EmbeddingGemma 2 delivers low-latency semantic representations for on-device applications, like search, retrieval-augmented generation (RAG), classification, and clustering.
Today, we’re launching EmbeddingGemma 2, expanding beyond text to unify code, images, video, and audio in a shared embedding space. Built on the Gemma 4 architecture and released under a commercially permissive Apache 2.0 license, EmbeddingGemma 2 has 740 million parameters, making it optimal for on-device inference. It can help find a specific video clip from a voice memo, or search through hours of audio recordings based on a text query, all processed by a single, natively multimodal model. Built from the same technology as Gemini Embedding models, EmbeddingGemma 2 is: [...] Optimized for on-device performance: Runs efficiently within tight resource constraints. With quantization, on a Google Pixel 11 Pro, EmbeddingGemma 2 requires as little as ~191MB active RAM for text-only weights and ~567MB for the full multimodal model. Extended context ready: Features an 8K token context window (4x larger than EmbeddingGemma 1), allowing it to process up to 5.5 minutes of audio, 29 images, 58
Comments230 Reposts161 2 days agoView post 8 ### Google launches EmbeddingGemma 2 open multimodal embedding model @Google We’re releasing EmbeddingGemma 2, our first natively multimodal open model engineered for on-device embeddings. Built on the Gemma 4 architecture and released under an Apache 2.0 license, it goes beyond text to unify images, video, audio, and code in a single embedding space. Views1.01M Comments118 Reposts258 2 days agoView post 9 ### Promethee: multiplayer gamified work app now available @Nlacombe\_ introducing Promethee, an app that turns your work into a multiplayer game, to perform. now available for everyone. Views863.7k Comments250 Reposts290 3 days agoView post 10 ### Charming: Personal apps for our group chats @mrmagan\_
Meet EmbeddingGemma 2, our first natively multimodal open model for on-device embeddings. It expands beyond text to unify code, images, audio,
Sundar Pichai / @sundarpichai: Introducing EmbeddingGemma 2, a new open multimodal model that sets the standard for on-device efficiency. - our first open, natively multimodal embedding model - handles text, code, image, video, and audio tasks within a lightweight, modular 740M parameter form factor - ideal for offline, privacy... Shubham Saboo / @saboo\_shubham\_: What...you can now do multimodal RAG on your phone, fully offline and free. EmbeddingGemma 2 does text search in ~191MB of RAM, or text + image + video + audio in ~567MB. Measured on a Pixel 11 Pro. Open weights under Apache 2.0 Ivan Fioravanti / @ivanfioravanti: 740M parameters of magic! We need to try EmbeddingGemma 2 ASAP! Bluesky: [...] Bluesky: @unsloth.ai: Google releases EmbeddingGemma 2, a new open model that runs locally on 0.5GB RAM. — The 740M parameter embedding model combines a 270M text model with vision (170M) + audio (300M). — Run & train the model via Unsloth. — GGUF: huggingface.co/unsloth
# Introducing EmbeddingGemma 2: An open model for natively multimodal embeddings ## Channel: Google for Developers (verified) 2.68M subscribers 361 likes ### Description 3,083 views Posted: 2026-10-06 Sebastian Russo, Ivan Llanos, and Sahil Dua from Google DeepMind introduce EmbeddingGemma 2, a lightweight open model that maps text, images, video, and audio into a single unified embedding space. At just 740 million parameters, EmbeddingGemma 2 is engineered for on-device use cases. Watch this video to discover: [...] ### Transcript [0:00] Today, we're super excited to introduce EmbeddingGemma 2, a lightweight, open model for natively multimodal embeddings. [0:08] - Traditionally, cross-modal retrieval required separate models, one for text, another for vision, another for audio. [0:16] EmbeddingGemma 2 changes that by mapping text, images, video, and audio, or any combination of them into a single unified embedding space. [0:26] This means a picture of a cat, the word cat, and the s
