How AI describes DeepMind Gemma

Gemma is a family of lightweight, open-weight AI models from Google, built from the same research used for Gemini and designed for multimodal text-and-image tasks. Gemma 3 models accept text and image inputs, output text, offer both pre-trained and instruction-tuned variants, support 140+ languages, and provide a large context window (128K tokens for large sizes and 32K for smaller sizes). They are intended for tasks like question answering, summarization, and reasoning, and their small footprint enables deployment on laptops, desktops, or private cloud infrastructure to democratize access to state-of-the-art AI.