Tuesday at the Google I/O 2025 event, the company also introduced its latest AI model, Gemma 3n. With this new model, we wanted something that would be simple enough to operate on any device – smartphone, laptop, or tablet. Gemma 3n runs comfortably on devices with fewer than 2GB of RAM. This overarching quality is what makes this tool so accessible to a less technical audience.
Gemma 3n looks similar in design and architecture to the Nano released earlier under the Gemini banner. It’s designed for high performance, able to manage audio, text, images, and video quickly and seamlessly. This broad applicability creates opportunities for developers to leverage the model in a myriad of different applications. Gemma models have been downloaded hundreds of thousands of times by developers, collecting tens of millions of downloads collectively. This accomplishment has come under fire for its proprietary, non-standard licensing terms.
“Gemma 3n shares the same architecture as Gemini Nano and is engineered for incredible performance,” said Gus Martins, a representative from Google. This feature makes Gemma 3n an attractive choice for developers wanting to build powerful applications on devices with limited resources.
Beyond Gemma 3n, Google made waves with its announcement that MedGemma would be released through its Health AI Developer Foundations program. MedGemma, meanwhile, does a better job than any other open Google model at understanding health-related text and images. It works really well out of the box for a range of image and text use cases, allowing developers to customize it for their own health-oriented applications.
Along with today’s announcement, Google made public its plans to build SignGemma, an open model to translate sign language into spoken-language text. “SignGemma is a new family of models trained to translate sign language to spoken-language text, but it’s best at American Sign Language and English,” Martins explained. He further emphasized the potential impact of SignGemma, stating, “It’s the most capable sign language understanding model ever, and we can’t wait for you — developers and deaf and hard-of-hearing communities — to take this foundation and build with it.”
Gemma 3n also goes into preview on Tuesday, so developers will have the perfect opportunity to experience its impressive capabilities. Some developers expressed concern that the license terms are too hazardous for commercial usage. The enthusiasm for the promise of these models remains remarkably high.