Model2h ago
Google DeepMind Releases Gemma 4 12B
The 12B parameter Gemma 4 model removes separate encoders to process text, images, and audio in a single stream. This unified architecture streamlines multimodal reasoning. It outperforms larger predecessors on complex visual tasks while maintaining a smaller footprint. Developers can now deploy more efficient, native multimodal pipelines without managing multiple disparate model components.