Google releases EmbeddingGemma 2 multimodal embedding model
What happened
Two reports on October 6 describe Google's EmbeddingGemma 2. One calls it an open-weight multimodal embedding model from Google DeepMind with 740M parameters, mapping text, images, video frames and audio into a single vector space and supporting local offline retrieval. The other says it builds on Gemma 4, ships under Apache 2.0, covers text, code, images, video and audio, and outputs unified 768-dimension vectors.
Written by AI from the coverage · updated 1 hour ago
Coverage
Follow the reports to see the story from different sides.
- AI HOT · ProductsGoogle DeepMind 发布多模态嵌入模型 EmbeddingGemma 2
Google DeepMind 发布开放权重多模态嵌入模型 EmbeddingGemma 2，以 740M 参数将文本、图像、视频帧和音频映射到统一向量空间，支持本地离线检索。
- AI HOT · Tips & opinionsGoogle 发布 EmbeddingGemma 2：基于 Gemma 4 的开源多模态嵌入模型
Google 发布基于 Gemma 4 的 EmbeddingGemma 2，以 Apache 2.0 许可开放，将文本、代码、图像、视频和音频映射到统一的 768 维向量空间。
Heat over time
Not enough continuous observations to draw a trend yet.