14h agoGoogle releases Gemma 4 12B, an open-source encoder-free multimodal model designed for local inferenceA1 top authorIt processes vision and text within a single decoder.Original postARArmand Joulin@armandjoulinRT @OlivierBachem: All parameters used for all modalities! Really excited and proud to release this competitive multimodal 12B model that…11:26 PM · Jun 3, 2026SentimentSentimentPos100%Neg0%Users are excited about Google's Gemma 4 12B encoder-free multimodal model release because of the team's collaborative work on the innovative training approach.2 comments with sentiment. View comments.1 more postRetweeted by Armand Joulin·4hView on
