.png){kind=logo}
Google has launched the latest addition to its family of ‘open’ AI models, Gemma 3n. It was unveiled during the Google I/O 2025 event. Designed with portability in mind, Gemma 3n is built to run smoothly on phones, tablets, and laptops, even those with under 2GB of RAM. The model is now available in preview starting today.
Google says Gemma 3n can process audio, text, images, and videos. It brings multimodal capabilities to edge devices without relying on cloud processing. This is a significant leap toward lightweight, on-device AI that is cost-effective and privacy-conscious.
“Gemma 3n shares the same architecture as Gemini Nano and is engineered for incredible performance,” said Gus Martins, Product Manager for Gemma, during the I/O keynote.
Offline AI models like Gemma 3n minimize data transfer, reduce latency, and maintain user privacy. They are specifically good in sensitive applications. By eliminating the need to send data to remote servers, Gemma 3n offers a more secure and efficient approach to mobile AI.
Alongside Gemma 3n, Google announced two more models expanding its AI ecosystem:
MedGemma: Developed under the Health AI Developer Foundations program, this model specializes in understanding medical text and images. Google calls it its most capable open model for healthcare-related multimodal analysis.
“MedGemma works great across a range of image and text applications,” Martins explained, encouraging developers to leverage it in health tech solutions.
SignGemma: Aiming to enhance accessibility, SignGemma is an open model. It is trained to translate sign language into spoken-language text. The Google AI model is particularly optimized for American Sign Language (ASL) and English.
“It’s the most capable sign language understanding model ever,” Martins stated. Further noting, “We can’t wait for developers and the deaf and hard-of-hearing communities to build with it.”
Despite its promise, Gemma has faced criticism for its custom, non-standard licensing terms. Still, that hasn’t weakened its appeal. Gemma models have been downloaded tens of millions of times, highlighting their growing adoption in the AI development community.
Google’s Gemma 3n represents a major step forward in making powerful, privacy-conscious AI accessible on everyday devices. Multimodal AI can now run efficiently offline on phones and tablets with limited resources. Google is setting a new standard for mobile AI innovation with this launch.
Also Read:Google I/O 2025 Preview: Android 16, Gemini AI, and Wear OS 6 Updates
