Google’s new Gemma 4 open AI model is sized for your laptop

In response to the soaring costs of memory driven by the generative AI revolution, Google is stepping up by introducing a new, less resource-intensive AI model. The latest addition to the Gemma 4 family, the Gemma 4 12B, is designed to run smoothly on standard consumer laptops, making advanced AI technology more accessible than ever. Earlier this year, Google launched four models within the Gemma 4 series, which marked a pivotal shift towards a more open Apache 2.0 licensing approach. The initial lineup included two mobile-focused models (E2B and E4B) along with two more robust options intended for serious computational tasks (26B Mixture of Experts and 31B Dense). However, there was a noticeable gap in the middle, which the new Gemma 4 12B model now fills. This new model boasts significant capabilities, offering a level of performance that surpasses the mobile options without requiring the hefty investment of a $20,000 AI accelerator. According to Google, the Gemma 4 12B can efficiently operate on any laptop equipped with at least 16GB of system RAM or VRAM. This requirement is approximately half that of the more demanding Gemma 4 26B MoE model, yet Google asserts that the new model performs nearly on par in benchmark tests, providing a compelling option for users seeking powerful AI solutions without breaking the bank.