ARTICLE AD BOX
New Delhi: Google has officially introduced its latest Gemma 4, the most capable version of its open-source AI model. The latest models also mark a significant step forward in making frontier-level AI accessible to developers everywhere, from powerful data center workstations to devices in users’ hands.
Google has also claimed that since it has released its first-generation Gemma, developers have downloaded it more than 400 million times, spawning a community ecosystem of over 100,000 model versions that are built on top of Google’s foundation.
Sundar Pichai, CEO of Google, has also stated that the models pack an incredible amount of intelligence per parameter, whereas Google DeepMind’s chief executive called Gemma 4 the greatest open models in the world of their respective sizes.
Sundar Pichai has also stated that, “Gemma 4 is officially here, the best open models in the world for its respective sizes. He also stated that it is available in 4 sizes that can be fine-tuned for users’ specific tasks: 31B dense for the great raw performance, 26B MoE for the low-latency, and effective 2B and 4B for the edge device use.”
The latest AI model is available under an Apache 2.0 license, which actually means the developers can use, modify, and build on it freely. The model can handle multi-step planning and complex logic, with the improvements in mathematics and instruction following tasks.
Gemma 4 natively supports function calling, structured data output, and system instructions by enabling developers to build AI agents that can interact with external tools, APIs, and services autonomously. Developers can also run Gemma 4 entirely offline on the local machine by turning a standard workstation into a private AI coding assistant.
All four models can process images and video natively; the two smaller edge models also support audio input for speech recognition. These edge models can process up to 128,000 tokens in a single prompt, while the larger models stretch to 256,000 tokens. Gemma 4 has been trained natively across more than 140 languages, making it one of the most globally inclusive open models available.
Google has actually made it while keeping it powerful. E2B and E4B models are built from the ground up in close collaboration with Google’s Pixel team, Qualcomm Technologies, and MediaTek, the companies behind the chips that power billions of Android devices worldwide. Model that runs completely offline, with near-zero latency, on everyday devices by including devices, Raspberry Pi boards, and Nvidia Jetson hardware.
English (US) ·