Wire flash
Google releases open-source multimodal embedding model EmbeddingGemma 2
Editorial responsibility
- No named human review is recorded for this page.
- Source reporting is collected, normalized, translated or condensed automatically when needed.
- Automatically published source-backed update
Google has announced the release of EmbeddingGemma 2, an open-source embedding model based on the Gemma 4 architecture. The model, distributed under the Apache 2.0 license, is designed to map multiple data modalities—including text, code, images, video, and audio—into a unified 768-dimensional vector space. The model's parameter count ranges from 270 million for text and code-only configurations to 740 million for the full multimodal version, allowing developers to load only the necessary components for their specific use case. The Google Developers Blog post provides detailed configuration options, dimension compression storage figures, and selection recommendations to help developers plan local multimodal retrieval solutions. This release aims to advance accessible multimodal AI capabilities for the developer community.
Source report
Google has launched EmbeddingGemma 2, an open-source embedding model built on the Gemma 4 architecture, released under the Apache 2.0 license. The model maps text, code, images, video, and audio into a unified 768-dimensional vector space.
Model Variants and Parameters
- Text/Code variant: 270M parameters
- Full multimodal variant: 740M parameters (supports text, code, images, video, and audio)
The model supports on-demand loading based on the required modality.
Technical Details and Recommendations
The original documentation provides specific storage figures and configuration parameters for each modality, along with dimensionality compression details. Developers can use this information to plan local multimodal retrieval solutions accordingly.
Source
aihotNeutral / independent
Part of this Story
Google releases EmbeddingGemma 2, an open multimodal AI model for on-device use