Google announced its new artificial intelligence model Gemma 3n, which can work directly on mobile devices, at the I/O 2025 event. The model can work on phones, laptops and tablets without any cloud connection.
Google announced the Gemma 3n artificial intelligence model
Gemma 3n is positioned as the newest member of Google’s Gemma family of open-source artificial intelligence models. According to the information shared by the company, this model is not limited to text data; it can also process multiple media types such as audio, visual and video.

One of the most striking details is that the model can work even on devices with only 2 GB of RAM. Google manager Gus Martins said that Gemma 3n is based on the same architecture as Gemini Nano.
Stating that the model works entirely on the device, Martins stated that this way, data does not need to be sent to the cloud and user privacy is protected. It was emphasized that working on the device also provides the advantage of avoiding high transaction fees.
Gemma 3n uses a new technique developed by Google DeepMind called “Per-Layer Embeddings” (PLE). This technology reduces the model’s memory usage, allowing it to run efficiently even on low-end devices.
Google is also ambitious about performance on mobile. According to shared data, Gemma 3n can respond 50 percent faster on mobile devices compared to the previously introduced Gemma 3 4B model.
So what do you think about this? You can share your opinions with us in the comments section below.