Google Unveils Gemma 3 AI Models Designed to Run on Single GPU

Google has launched Gemma 3, a new series of artificial intelligence models capable of running on a single GPU. The innovative Gemma 3 offers developers a portable solution for building AI applications across multiple devices, including phones, laptops, and workstations.

Built using the same advanced research and technology as Google's Gemini 2.0 models, Gemma 3 comes in multiple parameter sizes - 1B, 4B, 12B, and 27B - allowing users to select options that best match their hardware capabilities and performance requirements.

According to Google, the new AI models outperform competitors like Llama-405B, DeepSeek-V3, and o3-mini in preliminary human preference evaluations on LMArena's leaderboard. The company claims Gemma 3 supports over 35 languages with pre-trained capabilities spanning more than 140 languages.

Gemma 3 enables developers to build applications that analyze images, text, and short videos. The models feature a 128k-token context window, allowing them to simultaneously process much larger blocks of text. This expanded context window helps applications manage and interpret complex datasets, such as lengthy documents or multi-step processes.

"As the industry develops more powerful models, it will be critical to collectively develop risk-proportionate approaches to safety. We will continue to learn and refine our safety practices for open models over time," Google stated in its blog post.

The release of single-GPU capable AI models comes amid intensifying competition in the artificial intelligence sector, particularly following the growing popularity of DeepSeek, an AI model known for operating effectively on lower hardware requirements.

Meanwhile, other tech giants are making significant moves in the AI space. Meta has reportedly begun testing its first in-house chip for training AI models, aiming to reduce dependence on third-party suppliers like NVIDIA. Microsoft is allegedly exploring AI solutions from xAI, Meta, and DeepSeek as potential alternatives to OpenAI's technology for its Microsoft 365 Copilot product.