EmbeddingGemma 2: Unified and Efficient Multimodal Representations
Comments
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Summary
EmbeddingGemma 2 is an open 744M-parameter embedding model from Google that maps text, images, video, audio, and interleaved combinations into one shared vector space. Its modality encoders are modular, so it can run as text-only (271M), vision-language (439M), or audio-language (577M).