Google launched Gemma 4, a new family of open-weight AI models that the company says are its “most intelligent” models to date.
The April 2 release builds on growing adoption of the Gemma series, which has seen more than 400 million downloads and 100,000 community-built variants since its debut in February 2024.
Designed to deliver major advancements in reasoning, code generation and complex logic tasks, Google says the family is built using the same research and tech as Gemini 3 but marks a significant leap forward in capabilities.
Gemma 4 is available in four sizes to meet different environmental criteria. The smaller 2-billion- and 4-billion-parameter “Effective” models are intended for edge devices such as smartphones, while the 26-billion-parameter mixture-of-experts and 31-billion-parameter dense models can be deployed in more compute-intensive workloads.
Parameters are the settings a large language model can use to generate an output. While more parameters yield better results, they also require more compute to run. With Gemma 4, Google said it hopes to offer a more effective alternative to previous models, achieving “an unprecedented level of intelligence-per-parameter.”
“For developers, this new level of intelligence-per-parameter means achieving frontier-level capabilities with significantly less hardware overhead,” the vendor said in a blog post.
The models have been trained on more than 140 languages and feature built-in audio and visual processing, enabling them to run offline.
Despite their relatively modest size, Google claims the larger models deliver near-frontier performance, with the 26B variant outperforming models up to 20 times its size.
Accessibility is also a key differentiator with this release, with Gemma 4 distributed under an Apache 2.0 license (previous iterations were available through Google’s Gemma license). The change gives developers more freedom to modify and deploy the models commercially.
“This open source license provides a foundation for complete developer flexibility and digital sovereignty — granting you complete control over your data, infrastructure, and models,” Google said. “It allows you to build freely and deploy securely across any environment, whether on premises or in the cloud.”
