Google declared Gemma 4 to be the most intelligent open model family it has ever released on Thursday. Gemma 4 is the most recent generation of its open AI models, which are intended for sophisticated reasoning and agentic workflows. AzerNEWS reported, citing international media, that Gemma 4 is distributed under the Apache License 2.0 and is based on the same research and technological foundation as Gemini 3, according to a post on Google’s official blog.
According to Google, the new model family is intended to increase developers’ access to powerful AI capabilities. Gemma 4 will support a variety of hardware, including Android smartphones, laptop GPUs, developer workstations, and specialised accelerators. It will come in four sizes.
The models are appropriate for creating autonomous agents that can communicate with tools and application programming interfaces because they can perform sophisticated logical analysis and multi-step reasoning. Additionally, Google emphasised that all four Gemma 4 models can handle images and video natively and support code development, creating new opportunities for multimodal applications.
While bigger models can manage up to 256K tokens, enabling lengthy conversations or complicated job handling, edge models provide a 128K context window. Furthermore, Gemma 4 was trained on more than 140 languages, guaranteeing extensive multilingual support.
According to Gemma 4 AI, Gemma has been downloaded over 400 million times since its first release, and the “Gemmaverse” community has produced over 100,000 model variations, demonstrating the ecosystem’s explosive expansion and inventiveness. Notably, some developers are already experimenting with Gemma 4 in fields including intelligent workflow automation, AI-assisted coding, and real-time video analysis, suggesting that the paradigm could revolutionise both content creation and software development.
