Google DeepMind Senior Vice President Koray Kavukcuoglu announced that Gemini 4 could launch significantly before the end of the year. Speaking at The Information’s AI Agenda Live Summit, Kavukcuoglu confirmed the flagship model has entered the early post-training phase. This stage focuses on refining the base model to ensure safety and performance reliability.
Google intends to release an early version as soon as possible following promising initial results. The accelerated timeline supports Google’s strategy to gain a competitive edge against OpenAI, Anthropic, and Meta. This move signals an aggressive push to lead the rapidly evolving AI model market.