Google’s new Gemma 4 12B model is designed to run on any laptop with 16GB of RAM
Summary
Google announces Gemma 4 12B, a 12-billion-parameter AI model designed to run on consumer laptops with 16GB RAM. It uses new Multi-Token Prediction and a streamlined, native multimodal embedding approach to achieve near-parity with larger Gemma variants without requiring expensive accelerators. Weights are available for download via Kaggle and Hugging Face, and the model can run locally through tools like LM Studio and Google Edge Gallery under an open Apache 2.0 license.