What changed

Google released Gemma 4 12B weights under Apache 2.0 with text, vision, and native audio understanding, and says the model can run in roughly 16 GB of memory.

Why it matters

A compact, permissively licensed multimodal model expands practical local inference beyond text and images into audio.

What to do

Benchmark it locally for private transcription, image-and-document extraction, and low-latency multimodal classification before adopting a hosted model for those tasks.

Caveat

Apache 2.0 covers the weights, but real memory use depends on precision, context length, runtime, and modality; Google's benchmark results are not independent.