Google DeepMind released the 740-million-parameter EmbeddingGemma 2 under Apache 2.0, extending its earlier text embedding model.
Searching across formats
Google describes finding a video clip from a voice memo or searching audio with text. These are announced capabilities, not independent performance measurements.
Original source
This report summarises the source below. Analysis is labelled separately; product and research claims remain attributed to their source.
Read the original at Google DeepMind