Google calls Gemma 3 the most powerful AI model you can perform on a GPU

A little over a year later Take out two “open” models Gemma ai Built from the same technology behind its gemini, Google is Family update with Gemma 3. According to the blog post, these models are intended to be used by developers creating AI applications capable of executing wherever they are necessary, on everything, from a telephone to a workstation with a care of more than 35 languages, as well as the possibility of analyzing the text, images and short videos.

The company says it is the “best model of monocylator in the world”, outperforming competition from Facebook’s Llama, In depthAnd open for performances on a host with a single GPU, as well as optimized capacities to operate on the GPUs of Nvidia and the dedicated AI equipment. Gemma 3’s vision encoder is also upgraded, with a support for high resolution and non -square images, while the new Shieldgemma 2 image safety classifier is available for use to filter both the image input and the exit for the content classified as sexually explicit, dangerous or violent.

Last year, it was not clear how much interest there would be in a model like Gemma, however, The popularity of Deepseek and others shows that there is an interest in AI technology with lower material requirements. Despite its advanced capacity statements, Google also says: “The improved performance of Gemma 3 prompted specific evaluations focused on its abusive potential for the creation of harmful substances; Their results indicate a low risk level. »»

What is exactly an “open” or “open source” AI model remains a subject of debate, and with Gemma de Google, which has focused on the license of the company which restricts what people are allowed to use it, which has not changed with this new version. Google continues to promote Gemma with Google Cloud Credits, and the Gemma 3 academic program will allow university researchers to ask for $ 10,000 of credits to accelerate their research.