At its recent developer event, Google announced a new version of its AI model called Gemma—and this time, it’s small and light enough to run directly on smartphones.
What is Gemma?
Gemma is Google’s family of AI models that can do things like write text, answer questions, summarise documents, and more. The original versions of Gemma, launched last year, needed powerful hardware to run. But with the latest update, the smaller Gemma 2B version can now work on mobile devices, even low-cost ones like the Raspberry Pi.
This means apps using AI may no longer need to rely on Google’s servers or be connected to the internet all the time.
Why this matters?
Running AI on your device (instead of in the cloud) has several advantages:
- It’s faster, since it doesn’t need to send data back and forth to a server
- It offers better privacy, because your data stays on your phone
- It works offline, which is useful in areas with poor connectivity
For example, AI tools like voice assistants, translation apps, or even note-taking tools could all become more powerful and private with this technology.
Google’s rules for safe use
Even though the model is available for public use, Google has added certain limits to make sure it’s used responsibly. Developers can access the model freely, but they can’t use it to create harmful or misleading content.
To support this, Google also released an updated toolkit that helps people build safe and reliable AI apps. This step follows wider concerns about how open AI models are being used online.
Where can developers find it?
The new version of Gemma is available on several platforms that developers commonly use—Kaggle, Hugging Face, and Google Cloud’s Vertex AI, among others. That means it’s easy for researchers, students, and startups to download the model and experiment with it.
Google says the model is already being used in some universities and research projects, and this lighter version will expand its reach further.
The trend toward on-device AI
Google isn’t the only tech company moving in this direction. There’s a growing trend to bring AI models to phones and smaller devices. It’s not just about saving power or internet—it’s also about giving users more control over their data and making AI tools faster and more reliable.
