The family of "open" AI models from Google, known as Gemma, is continuously expanding.

During the 2025 Google I/O conference on Tuesday, Google introduced Gemma 3n, a model specifically designed to run "smoothly" on mobile phones, laptops, and tablets. A preview version has been available starting from Tuesday. According to Google, Gemma 3n can handle audio, text, images, and video.

In recent years, within the AI community, efficient models that can operate offline without relying on cloud computing have become increasingly popular. Not only are they more cost-effective compared to large models, but they also protect privacy by eliminating the need to transmit data to remote data centers.

In addition to Gemma 3n, Google also released MedGemma through its Health AI Developer Base initiative. According to the company, MedGemma is its most powerful open model for analyzing health-related texts and images.

Another upcoming model is SignGemma, an open model that translates sign language into spoken text. Google stated that SignGemma will enable developers to create new applications and integrations for deaf and hearing-impaired users.

It's worth noting that Gemma has faced criticism due to its customized, non-standard licensing terms. Some developers have expressed that this makes commercial use of these models risky. However, this hasn't dampened the enthusiasm of developers, who have downloaded the Gemma models tens of millions of times.