Google has announced the release of the Gemma 3 AI model family, designed for developers building artificial intelligence applications. The new models support over 35 languages and can analyze text, images, and short videos, expanding the capabilities of multimodal applications.
Gemma 3 continues the evolution of the Gemma series, first introduced by Google in early 2024. These models offer high performance on single GPUs and are optimized for use on Google Cloud servers, as well as Nvidia GPUs and specialized AI accelerators for local deployments.
The updated version features an enhanced visual encoder, enabling high-resolution image processing and support for non-standard formats. Additionally, Google has introduced ShieldGemma 2, a security system that filters out images containing potentially harmful or undesirable content.
To support developers, Google is offering free credits for using Gemma models on Google Cloud. The company has also launched the Gemma 3 Academic Program, which provides up to $10,000 in funding for researchers working on AI projects.
Gemma 3 is now available to developers worldwide, offering powerful tools for advancing artificial intelligence applications.