Optimized on-device performance
Engineered for speed and quality, with a significantly reduced memory footprint.
Gemma 3n was created in close collaboration with leading mobile hardware manufacturers. It shares architecture with the next generation of Gemini Nano to empower a new wave of intelligent, on-device applications.
Slide 1 of 4
Privacy-first, offline-ready
Enables developers to build intelligent, interactive features that respect user privacy and work reliably offline.
Multimodal understanding
Understands and processes audio, text, images, and videos, and is capable of both transcription and translation.
Dynamic resource usage
Features a 4B active memory footprint with nested 2B active memory submodel – with the ability to create submodels for quality-latency tradeoffs.
Live interactive applications
Create apps that understand and respond to real-time visual and audio cues from the user's environment.
Applications based on deep understanding
Using combined audio, image, video, and text inputs—all processed privately on-device.
Advanced audio-centric applications
Including real-time speech transcription, translation, and rich voice-driven interactions.
Gemini API
Run Gemma with the Gemini API
Google AI Edge
Run large language models (LLMs) completely on-device
Hugging Face
Ollama
Kaggle