Google is preparing to launch its next flagship AI model, Gemini 4, which is currently in the refinement phase. Koray Kavukcuoglu, the new leader of the DeepMind division, stated that the company aims to release the model well before the end of the year. The team plans to provide an early post-training output as soon as possible to maintain a faster pace of development.
The company has not launched a major flagship model since the Gemini 3 series in November 2025. During this period, competitors such as OpenAI and Anthropic released their own models, GPT-6 and the Mythos series, which currently outperform the Gemini 3 lineup. Google previously considered a Gemini 3.5 Pro update in June, but the release was canceled to prioritize the development of faster Flash models.
This shift in strategy follows significant leadership changes at DeepMind, including the departure of former AI head Demis Hassabis in August. Despite the current competitive gap, Kavukcuoglu emphasized that Google remains committed to maintaining its position at the frontier of artificial intelligence development.
