Just before the launch of Gemma 4 AI, the tech world was buzzing with anticipation. Google DeepMind was set to unveil a family of state-of-the-art open models that promised to change the landscape of artificial intelligence.
On launch day, Gemma 4 made headlines by supporting over 140 languages, making it one of the most versatile AI models available. This significant feature allows developers and users worldwide to interact with AI in their native languages.
Gemma 4 is available under the Apache 2.0 license, ensuring that developers can freely use and modify the models. This open approach is expected to foster innovation and collaboration in the AI community.
Among its standout features, Gemma 4 enables multi-step planning, autonomous action, offline code generation, and audio-visual processing. These capabilities position it as a powerful tool for developers looking to create advanced applications.
The models can run on a variety of devices, including mobile, desktop, IoT, and robotics, making them accessible for a wide range of use cases. Notably, the Gemma 4 models feature a 128K context window, allowing for the processing of long-form content efficiently.
For those interested in speech recognition, the E2B and E4B models support native audio input. This feature enhances user interaction, making it easier to integrate voice commands into applications.
Performance-wise, Gemma 4 can achieve a prefill throughput of 133 tokens per second on a Raspberry Pi 5, showcasing its efficiency even on lower-end hardware.
The models are optimized for fine-tuning on a wide range of hardware, from billions of Android devices to developer workstations. This flexibility is crucial for developers aiming to tailor AI solutions to specific needs.
With 26B and 31B versions available, Gemma 4 is designed to cater to different performance requirements. The 26B Mixture of Experts model operates with 3.8 billion active parameters during inference, ensuring robust performance.
As the launch unfolds, the excitement is palpable. “The era of agentic experiences on-device is here, and we hope you are excited to start building on the edge,” a spokesperson from Google DeepMind remarked.
Gemma 4 is set to power the next generation of pioneering research and products, with its models specifically sized for efficient operation on various hardware. This development marks a significant step forward in the evolution of AI technology.
