Google DeepMind’s open-weight Gemma family has crossed one billion cumulative downloads, a milestone the company revealed on August 20.
The number is the first cumulative total Google has published for Gemma since the models debuted two years ago. DeepMind vice president Clement Farabet and product director Olivier Lacombe said in the announcement that outside developers have also shipped more than 100,000 variants built on the open weights.
Those variants have carried Gemma into coding, medicine, finance, and edge deployments, giving developers a customizable base that avoids the cost of frontier-scale training. The family sits at the open end of Google’s model lineup, with its flagship Gemini systems remaining closed.
The milestone gives Google a headline number in the open-model race, where rivals measure adoption in downloads, forks, and derivative models.
