According to two sources, Google co-founder Sergey Brin has been pushing the company's key artificial intelligence (AI) team to focus on the Gemini model in recent months, as parent company Alphabet works to close the gap with competitors. One anonymous source revealed that in April, when Anthropic took the lead with a preview of its powerful Claude Mythos model, Brin spoke at an internal meeting attended by hundreds of employees, urging Google DeepMind's AI lab to speed up development.

Last November, Google's Gemini model briefly surpassed rivals, but subsequent new versions from Anthropic and OpenAI put Google back in a catch-up position. By August, due to internal tests showing Gemini still lagging behind competitors in areas like coding, Google delayed the launch of its next-generation flagship Gemini by two months, according to one source and another person familiar with the matter.

On August 5, Google announced a major restructuring of the leadership at its Google DeepMind AI unit. Unit head Demis Hassabis stepped down from his management role, with his deputy Koray Kavukcuoglu taking over. Hassabis transitioned to a chairman position with reduced authority. Meanwhile, Gemini's two original technical co-leads resigned to start a new company.

While some industry observers question whether Google is stepping back from the large model race, interviews with seven people familiar with Gemini's development indicate that Alphabet's commercialization strategy and last week's personnel reshuffle are closely tied to the effort to maintain model dominance.