Felixnews.com – Google has officially unveiled Gemma 4, a next-generation open-source artificial intelligence (AI) model designed to run locally on devices such as Android phones and computers. As reported by infobae.com, this launch underscores Google’s commitment to development flexibility.
The model features 31 billion parameters, making it one of the most advanced open-source AI models available today. Gemma 4 is said to possess advanced reasoning capabilities and support agent-based workflows.
One of Gemma 4’s key advantages is that it is distributed under the Apache 2.0 license. This license permits commercial use of the model, giving developers complete freedom to modify and deploy the AI. Google emphasizes this as the foundation of digital sovereignty and absolute control over their data, infrastructure, and models.
Gemma 4 is available in four different sizes: 2 billion effective parameters, 4 billion effective parameters, 26 billion Expert Mix, and 31 billion Compact. These variants provide a context window of up to 256,000 tokens, allowing for optimal adaptation to various needs and hardware capacities.
Multimodal Capabilities and Local Processing
This model is capable of natively processing audio and video, and can generate code without requiring an internet connection, thanks to its focus on local processing. Additionally, Gemma 4 supports over 140 languages, expanding its global reach.
The model also offers direct JSON output, which is ideal for pipelines and agents without the need for complex post-processing. Its multimodal capabilities make it a bridge for solutions involving voice, images, and text, expanding the boundaries of creativity in startup development.