--- title: "What is Gemma from Google?" description: "What is Gemma from Google?" author: "Anna Weber" published: 2026-11-12 updated: 2026-11-12 canonical: https://answers.mindstick.com/qa/119994/what-is-gemma-from-google category: "Open-Source AI Models" tags: ["Gemma", "google", "open models"] reading_time: 2 minutes --- # What is Gemma from Google? ## Answers ### Answer by Sagar Poudel Gemma is a family of open-weight AI models released by Google DeepMind. They are built using the same research and technology as Google's Gemini models, but they are smaller and released so that developers can download, run and customise them on their own hardware. Gemma and Gemini are related but different. Gemini is Google's flagship model family, available through the Gemini app and Google's APIs, and its weights are not released publicly. Gemma consists of lighter models whose weights are available for anyone to download, designed to run efficiently on laptops, workstations, phones and modest servers. Google first released Gemma in February 2024 and has released several newer generations since, with improved quality, multilingual support, image understanding and longer context windows. The family includes models of various sizes, from very small versions suitable for phones to larger ones that rival much bigger models on many tasks. There are also specialised variants built on Gemma, for example for coding, for medical applications, for safety classification and for other research uses. Why do developers choose Gemma? 1. Efficiency. Gemma models are designed to perform well relative to their size, which makes them good for limited hardware.\ 2. On-device AI. Smaller versions can run on phones and edge devices without an internet connection.\ 3. Google's tooling. Good support across Google's AI tools, Google Cloud, Hugging Face, Ollama, LM Studio and other popular frameworks.\ 4. Multilingual support across many languages, including Indian languages to varying degrees.\ 5. Fine-tuning. Gemma is easy to adapt for specific tasks with techniques like LoRA. Common uses include private chatbots, document summarisation on local machines, mobile app features that work offline, educational projects and research. On licensing, Gemma models are released under Google's own terms of use, which allow commercial use but include a prohibited use policy. Read the terms for the specific version before building a product. If you want to try Gemma, the easiest way is through Ollama or LM Studio on your computer, or through Google AI Studio and Hugging Face. Start with a smaller version to see how it performs on your hardware and your tasks. For developers who want Google-quality AI that runs anywhere, including offline, Gemma is one of the strongest options among open models. --- Original Source: https://answers.mindstick.com/qa/119994/what-is-gemma-from-google Copyright © MindStick Software Pvt. Ltd. This Markdown version is provided for developers, AI systems, and offline reading.