Google has officially released Gemma 4, its next-generation open-source artificial intelligence model. According to Google, the lightweight versions of Gemma 4 demonstrate breakthrough efficiency, outperforming competitor models that are up to 20 times larger in scale.

Key Highlights of Gemma 4
Built on the same core technology as Google’s flagship Gemini 3, Gemma 4 represents the pinnacle of performance in Google’s open-source lineup. The release includes four distinct sizes to cater to different hardware requirements:

Lightweight Versions: 2B (2 billion) and 4B (4 billion) parameters.
Large-Scale Versions: 26B (26 billion) and 31B (31 billion) parameters.

Frontier Capabilities on the Edge
Moving beyond simple chat interfaces, Gemma 4 is designed to support agent-based workflows with native tool integration. Its most striking feature is the built-in multimodal capability (vision and audio processing), allowing the models to run offline on mobile and IoT devices with near-zero latency.

In terms of raw performance, the Gemma 4 31B model secured 3rd place among all open models worldwide on the industry-standard Arena AI Text Leaderboard, while the 26B version ranked 6th. This allows developers to access frontier-level reasoning and multimodal power without the burden of massive hardware costs.

 
Developer-Centric Ecosystem
To foster a robust developer community, Google has released the models under the Apache 2.0 license, facilitating easy commercial use. The models also adhere to Google DeepMind’s rigorous Gemini safety protocols, ensuring high standards of transparency and security.

"Gemma 4 provides the efficiency and fine-tuning flexibility developers need for the evolving 'Physical AI' ecosystem, where models must operate on the edge without constant internet connectivity."

Analysis: A Strategy for Ecosystem Dominance
Industry analysts view this release as a strategic move to dominate the AI ecosystem. By providing models that are small enough to run on edge devices but powerful enough to rival giant LLMs, Google is positioning itself as the primary infrastructure provider for the next wave of localized, autonomous AI applications.

[ⓒ Global Economic Times. 무단전재-재배포 금지]