Google's Gemma 4 open models deliver frontier AI performance on a single Nvidia GPU, with Apache 2.0 licensing and native support for agentic workflows.