Summary

Google DeepMind's Gemma 4 is a new family of open, developer-friendly AI models ranging from 2 to 32 billion parameters that can run on various devices from smartphones to consumer GPUs. The models offer multimodal capabilities, reasoning skills, and on-device intelligence across different sizes, with the smallest models capable of running on mobile phones and the largest providing high-level intelligent performance. These open models are designed to be accessible, allowing developers to download, run, and fine-tune them for specific use cases, marking a significant advancement in AI model accessibility and performance.