AsianFin – Google announced on Wednesday the launch of an upgraded version of Gemma, its open-source large language model (LLM) series.

According to Google's official blog, Gemma 3 is designed to help developers build AI applications that can run across multiple devices. The model supports over 35 languages and is capable of analyzing text, images, and short videos.

Google claims Gemma 3 is "the world's best single-accelerator model," outperforming competitors such as Facebook’s Llama, DeepSeek, and OpenAI when deployed on a single GPU. The upgrade aims to enhance accessibility and performance for AI developers working on diverse computing platforms.