Google Gemma 3n – Mobile AI Model with Multimodal Power
Google DeepMind has released Gemma 3n, an AI model built for phones, tablets, and lightweight laptops. It runs well on only 2–3 GB of memory, so developers can keep everything on the device and skip constant cloud calls.
Gemma 3n comes in two sizes: E2B (5 billion parameters) and E4B (8 billion parameters). Despite the large parameter count, it uses smart memory-saving techniques like Per-Layer Embeddings and int4 quantization to lower the runtime footprint. This means it performs like a much smaller model, making it suitable for edge devices.
A key innovation in Gemma 3n is the MatFormer architecture. This allows a smaller sub-model to run inside the larger one. Developers can switch between compact and full models depending on the task, without loading multiple models.
Gemma 3n already supports text and image inputs, and it’s built to handle audio and video as well. It uses a SigLIP vision encoder and a “Pan & Scan” method that helps process images of any shape or resolution efficiently.
It’s also fast, 1.5 times quicker than the previous Gemma 3 4B model. All processing is done locally, offering privacy and quick response even in areas without internet.
Developers can get started now using Google AI Studio, AI Edge Gallery, Hugging Face, or Vertex AI. Model weights are open and compatible with frameworks like Transformers and llama.cpp.
Gemma 3n is already being used in translation tools, healthcare apps, educational platforms, and more. With its support for offline use, low memory needs, and multimodal input, it sets a new standard for on-device AI. More features, including audio and video processing, are coming soon. For more details about visit the official site.
Stay Updated with the Latest news by Joining our Telegram and WhatsApp Channels.
Also Read:
- Higgsfield Soul an AI Photo Model with Realism and Ease
- ElevenLabs Mobile App Now Available for iOS and Android
- Perplexity AI Video Feature Launches on X – Know How It Works
- Higgsfield Speak Launched – AI Video Creation Now Just a Few Clicks Away
- Minimax Hailuo 02 – A Smart Step Forward in AI Video Creation
