Multimodal
6 stories in this blend
Superintelligence·4d agoGoogle DeepMind launches EmbeddingGemma 2 for local multimodal search
EmbeddingGemma 2 is a lightweight model capable of converting text, code, audio, and video into searchable vectors directly on mobile devices. It allows applications to process complex media queries locally without sending sensitive user data to external servers.
The Neuron·5d agoEmbeddingGemma 2
A lightweight Google search model that runs locally to process text, audio, images, and video. It suits developers building private, offline search experiences into mobile apps and web platforms.
The Neuron·Sep 25Gemini 3.8 Live
A voice model update from Google that adds real time spoken conversations in dozens of languages alongside animated digital avatars. It allows users to combine live camera feeds and audio input during interactive sessions.
The Neuron·Sep 23Xiaomi MiMo-V2.6
An open-weights artificial intelligence model family distributed under the MIT license for multimodal and multi-agent workflows. It provides a flagship Pro version alongside faster variants designed for developer experimentation.
AI Daily Brief·Sep 19Gemini 3.8 Live
Google introduced a real time speech model that supports continuous conversation, dynamic visual processing, and multi language recognition. It allows continuous back and forth interactions and can execute automated background tasks during active calls. It is designed for developers adding conversational voice interfaces into specialized software and hardware.
There's An AI For That·Aug 27Google updates video generation model to support longer high definition output
Google released an update to its Gemini Omni video model that extends context awareness up to 10 seconds. The model can fill in transitions between starting and ending keyframes to output 4K footage up to 40 seconds long.
