Gemma 4 just dropped and this one actually matters.

31B model, Apache 2.0, reasoning that competes with models 10x its size. You can run this locally, wire it into your own automation pipelines, no API costs, no rate limits, no vendor lock-in.
As someone who builds production-grade systems at scale — this is the open-source moment I've been waiting for.
What are you building with it? 👇