Tech giant Google has launched its new open source AI model series Gemma 4, which can now work everywhere from data centers to smartphones. The company claims that this model gives very powerful performance even in small size. CEO Sundar Pichai and DeepMind chief Demis Hassabis have called it a game changer for developers. The special thing is that this AI can work on the device even without internet.
Google has introduced Gemma 4 in four different sizes to meet all types of devices and needs. It has E2B and E4B smaller models, which are specially designed for smartphones and IoT devices. Whereas 26B Mixture of Experts model gives mid level performance and 31B Dense model is the most powerful version. According to the company, the 31B model has left behind many big AI models, even though its size is much smaller than them. This gives developers the freedom to choose the right model according to their project.
Gemma 4 has emerged as not just a chatbot but an advanced AI platform. It can understand multi step reasoning, maths and complex questions better. Additionally, it has agentic workflow support, through which AI itself can interact with tools and APIs. With the help of code generation, developers can also use it as an offline coding assistant. Apart from this, it can also understand image, video and audio input and works in more than 140 languages, making it very useful for global use.
The biggest feature of Gemma 4 is that it runs on device. Google has designed it in collaboration with companies like Qualcomm and MediaTek, so that it can work easily on billions of Android devices. This model runs completely offline, giving users a fast and secure AI experience. According to the company, it has been developed in collaboration with the Pixel team, which opens up new possibilities of AI on smartphones. This technology can make mobiles more smart and self-reliant in the future.
