Models · The Decoder ·

Google Deepmind's Gemma 4 12B squeezes multimodal AI onto a laptop with just 16 GB of RAM

Google Deepmind's Gemma 4 12B squeezes multimodal AI onto a laptop with just 16 GB of RAM

Google DeepMind's Gemma 4 12B is an open-source multimodal model that natively handles text, images, and audio. The company says it can run on laptops with 16 GB of RAM and performs close to the larger 26B version, with Apache 2.0 licensing for commercial use.

Read the full story at The Decoder →