Google has introduced Gemma, a set of “open models” for making AI software, making advanced AI tech more accessible to everyone. With Gemma, developers, and businesses can now access model weights and technical data for free, removing the barrier of high costs. This move supports innovation and helps Google expand its presence in cloud computing.
Gemma models are designed for Google Cloud, and developers get $300 in credits for trying it out, which encourages more people to use Google Cloud services and can increase Google’s revenue.
Google’s Gemma isn’t completely open source, but it’s a compromise between keeping control and allowing collaboration. Google wants to prevent misuse while still letting the AI community benefit from contributions.
Unlike Meta’s Llama 2 models, which can be very large, Gemma models are smaller, with two billion or seven billion parameters. While Google hasn’t said how big its largest Gemma models are, they’re perfect for developers who don’t need huge models but still want powerful AI.
Google is also working with Nvidia, a chipmaker, to make Gemma models work well with Nvidia’s chips. This means Gemma can run on a wider range of hardware. Nvidia is even making chatbot software compatible with Gemma, making it easier to use on different computers.
Google’s introduction of Gemma is a big move towards making AI technology more accessible to everyone. It encourages innovation and makes it easier for businesses to use cloud services. Google wants to be a top player in AI and cloud computing by finding the right balance between control and openness.