Google calls Gemma 3 the most powerful AI model you can run on one GPU

A little over a year after releasing two “open” Gemma AI models built from the same technology behind its Gemini AI, Google is updating the family with Gemma 3. According to the blog post, these models are intended for use by developers creating AI applications capable of running wherever they’re needed, on anything from a phone to a workstation with support for over 35 languages, as well as the ability to analyze text, images, and short videos.

The company claims that it’s the “world’s best single-accelerator model,” outperforming competition from Facebook’s Llama, DeepSeek, and OpenAI for performance on a host with a single GPU, as well as optimized capabilities for running on Nvidia’s GPUs and dedicated AI hardware. Gemma 3’s vision encoder is also upgraded, with support for high-res and non-square images, while the new ShieldGemma 2 image safety classifier is available for use to filter both image input and output for content classified as sexually explicit, dangerous, or violent.

Last year it was unclear how much interest there would be in a model like Gemma, however, the popularity of DeepSeek and others shows there is interest in AI tech with lower hardware requirements. Despite its claims of advanced capabilities, Google also says, “Gemma 3’s enhanced STEM performance prompted specific evaluations focused on its potential for misuse in creating harmful substances; their results indicate a low risk level.”

What exactly constitutes an “open” or “open source” AI model remains a topic of debate, and with Google’s Gemma, that has focused on the company’s license that restricts what people are allowed to use it for, which has not changed with this new release. Google is continuing to promote Gemma with Google Cloud credits, and the Gemma 3 Academic program will allow academic researchers to apply for $10,000 worth of credits to accelerate their research.