Science - Technology

Google launches 'super lightweight' AI model

TB (summary) • March 13, 2025 20:13

Google has just introduced Gemma 3, a third-generation open-source AI model that can run on both smartphones and high-performance workstations.

The open source Gemma 3 model stands out for its ability to run on a single GPU.

Google says it's a modern, lightweight open source framework built from the same research and technology as Gemini 2.0. Gemma 3 comes more than a year after the company first announced the open source Gemma framework.

Gemma 3 is for developers creating AI applications that can run software anywhere, on anything from phones to workstations, with native support for over 35 languages, the ability to analyze text, images, and short videos.

Gemma 3 will be available in 1B, 4B, 12B and 27B versions, allowing users to make the choice that best suits their specific hardware and performance needs.

Mô hình Gemma 3 chỉ thua kém R1 của DeepSeek trong bảng xếp hạng của LMArena. Ảnh: Seekingalpha — Gemma 3 model is only behind DeepSeek's R1 in LMArena's rankings

Google also highlighted that Gemma 3 outperformed Meta Platforms Llama-405B, DeepSeek V3, and OpenAI's o3-mini in early evaluations on LMArena. In this ranking, only Deepseek's R1 model scored higher than Google's new AI model.

Gemma 3 is available in 35 languages and has trained support for over 140 languages.

Google also launched ShieldGemma 2, a 4B image safety checker, built on Gemma 3.

TB (summary)