Introducing Gemma 4 12B: a unified, encoder-free multimodal model
Google DeepMind announces Gemma 4 12B, a unified encoder-free multimodal open model.
Google DeepMind introduced Gemma 4 12B, described as a unified, encoder-free multimodal model, signaling continued momentum in the open-weight model space from a major lab. A new architecture approach (encoder-free multimodality) at this parameter scale is a notable competitive move in the open model ecosystem. Note: only the announcement title was available—no article body or quotes were provided to analyze.