Just before the launch of Gemma 4 AI, the landscape of artificial intelligence was rapidly evolving, with developers seeking more efficient and versatile tools. On October 2023, Google DeepMind unveiled Gemma 4, a family of state-of-the-art open models designed to enhance AI capabilities across various platforms.
Gemma 4 supports over 140 languages, making it a highly accessible tool for developers worldwide. This model is available under the Apache 2.0 license, promoting open-source collaboration and innovation.
One of the standout features of Gemma 4 is its ability to enable multi-step planning and autonomous action, which is crucial for developing sophisticated AI applications. Additionally, it supports high-quality offline code generation, allowing developers to turn their workstations into local-first AI code assistants.
As of now, Gemma 4 models can run on a variety of devices, including mobile phones, desktops, IoT devices, and robotics, demonstrating its versatility. The models feature a 128K context window, which is particularly beneficial for processing long-form content.
The E2B and E4B models of Gemma 4 also support native audio input for speech recognition, enhancing user interaction. Impressively, 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.
Moreover, the models are optimized for efficient fine-tuning on hardware ranging from billions of Android devices to developer workstations. The Gemma 4 family includes versions with 26B and 31B parameters, specifically tailored for different hardware capabilities.
Developers are particularly excited about the potential of Gemma 4 to build autonomous agents that can interact with various tools and APIs, marking a significant step forward in AI development. As one spokesperson noted, “The era of agentic experiences on-device is here, and we hope you are excited to start building on the edge.”
With these advancements, Gemma 4 is set to power the next generation of pioneering research and products, as it is sized specifically for efficient operation on a wide range of hardware.
As the technology continues to evolve, the implications for developers and researchers are profound, offering new opportunities for innovation in AI applications.
