FrontierThe story, in brief
Google Deepmind's Gemma 4 12B squeezes multimodal AI onto a laptop with just 16 GB of RAM
Multimodal AI that runs on 16GB laptops. Google DeepMind's Gemma 4 12B just changed the efficiency game.
The KeyNews take
Why it matters
Google DeepMind's Gemma 4 12B demonstrates a major capability-per-compute breakthrough: native multimodal processing (text, image, audio) at 12B parameters with performance parity to larger models, dramatically lowering the barrier to deployment and expanding the addressable market for on-device AI.
The key facts
5 to knowGemma 4 12B runs on laptops with 16GB RAM
Multimodal: native text, image, and audio processing
Performance nearly matches 26B model on benchmarks
Apache 2.0 license enables commercial use
Open-source release
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Google Deepmind's Gemma 4 12B is an open-source model that processes text, images, and audio natively and runs on laptops with just 16 GB of RAM. It nearly matches the twice-as-large 26B model in benchmarks and ships under an Apache 2.0 license for commercial use.