March 12 News.GoogleLaunched today Gemma 3 Artificial Intelligence model, which follows its release a year ago based on a partnership with the Gemini AI The upgrade comes on the heels of two "open" Gemma AI models of the same technology. According to Google's official blog, the Gemma 3 model is designed for developers to build AI apps that run seamlessly on multiple devices, whether they are cell phones or workstations. The model supports more than 35 languages and has the ability to analyze text, images and short videos.
Google claims that Gemma 3 is "the world's best single-accelerator model," outperforming competitors like Facebook's Llama, DeepSeek, and OpenAI on hosts with a single GPU. The model is also optimized for Nvidia's GPUs and dedicated AI hardware. Google has also released a 26-page technical report that goes into depth on these performance benefits.
Over the past year, the popularity of models such as DeepSeek has demonstrated the market's strong interest in AI technologies with lower hardware requirements, despite doubts about whether models like Gemma are sufficiently attractive. Despite Gemma 3's claims of advanced capabilities, Google has also said that the model's improved performance in science, technology, engineering, and math (STEM) has triggered a specific assessment of its potential use in the manufacture of hazardous materials, but that the assessment shows a low level of risk.
1AI notes that the definition of "open" or "open source" AI models is still a matter of debate, and Google's Gemma model is limited in its use by the company's license, which has not changed with the release of Gemma 3. Google continues to promote Gemma models through Google Cloud Credits and has launched the Gemma 3 Academic Program, which allows academic researchers to apply for $10,000 worth of credits to accelerate their research.