Google’s publicly accessible Gemma AI model suite has surpassed 150 million downloads, marking a significant adoption milestone for the company’s open-source initiative
Over the weekend, Omar Sanseviero, a developer relations engineer at Google DeepMind, disclosed the figure on X.
Additionally, he disclosed that developers have generated over 70,000 variations of Gemma on the AI development platform Hugging Face.
In February 2024, Google introduced Gemma with the objective of competing with other “open” model families, such as Meta’s Llama.
The most recent Gemma releases are multimodal, which means they can operate with both text and images, and they support more than 100 languages.
Additionally, Google has developed variants of Gemma that are specifically designed for specific applications, such as drug discovery.
Although 140 million downloads in approximately one year is a remarkable statistic, Gemma is significantly behind Llama, its most significant competitor, which exceeded 1.2 billion downloads in late April.
It is important to mention that Gemma and Llama have been criticized for their custom, non-standard licensing terms.
According to some developers, these terms have made the commercial use of the models a hazardous endeavor.
