Azernews.Az

by Alimat Aliyeva

On Thursday, Google announced Gemma 4, the latest generation of its open AI models designed for advanced reasoning and agentic workflows, calling it the most intelligent open model family it has released to date. According to a post on Google's official blog, Gemma 4 is built on the same research and technology foundation as Gemini 3 and is released under the Apache License 2.0, AzerNEWS reports, citing foreign media.

Google stated that the new model family aims to make advanced AI capabilities more widely accessible to developers. Gemma 4 will be available in four sizes, supporting a range of hardware—from Android devices and laptop GPUs to developer workstations and specialized accelerators.

The models are capable of multi-step reasoning and deep logical analysis, making them suitable for building autonomous agents that can interact with tools and application programming interfaces. Google also highlighted that all four Gemma 4 models support code generation and can natively process images and video, opening new possibilities for multimodal applications.

Edge models offer a 128K context window, while larger models can handle up to 256K tokens, enabling extended conversations or complex task handling. Additionally, Gemma 4 was trained on over 140 languages, ensuring broad multilingual support.

Since the launch of its first generation, Google reports that Gemma has been downloaded more than 400 million times, and the community known as the “Gemmaverse” has created over 100,000 model variants, showcasing the ecosystem's rapid growth and creativity. Notably, some developers are already experimenting with Gemma 4 in areas like real-time video analysis, AI-assisted coding, and intelligent workflow automation—hinting at the model’s potential to reshape both software development and content creation.