Google releases Gemma 4 12B for 16GB laptops

120亿参数跑在16G笔记本上，谷歌Gemma 4新成员杀来了

Google released Gemma 4 12B, a medium-size model that runs locally with 16GB VRAM or unified memory. It uses an encoder-free multimodal architecture, supports native audio input, ships under Apache 2.0, and includes an MTP draft model for lower latency.

Why it matters: Google’s Gemma 4 12B has clear HKR-H/K/R: 16GB local running, 12B scale, and Apache 2.0 licensing. It is a strong open-model update, not a must-write foundation-model launch.