Google DeepMind chief Koray Kavukcuoglu said in an interview that the company's next flagship model, Gemini 4, is in its refinement stage and that Google is aiming to launch it "much earlier" than the end of 2026. It was Kavukcuoglu's first media appearance since taking over the DeepMind division, following a leadership shakeup in August when former AI head Demis Hassabis stepped down after saying "great progress" was being made on Gemini 4.
Google has not shipped a new flagship model since the Gemini 3 series arrived in November 2025. In the meantime, OpenAI released GPT-6 and Anthropic launched its Mythos series, and both outperform Gemini 3. At Google I/O in May, CEO Sundar Pichai said a major Gemini 3.5 Pro update would arrive in June to help close the gap, but that model was never rolled out.
Kavukcuoglu acknowledged the pause, saying the team "took a little bit of a step back" to prioritize faster, less powerful Flash models instead of shipping Gemini 3.5. Google now wants to put out an early post-training output of Gemini 4 "as soon as possible" because the team is excited by the results, he said, and will "continue the fast-paced iterations" afterward.
Why it matters
GPT-6 and Mythos give OpenAI and Anthropic a sizable lead over Google's current flagship. A Gemini 4 release before year-end would be Google's first new flagship since November 2025, and the first real test of whether the reorganized DeepMind leadership can keep up the faster cadence Kavukcuoglu described.
Our take
"Much earlier" is not a date, and Google already missed one public target this year when the Gemini 3.5 Pro update promised for June never shipped. The more telling detail is the plan itself: release an early post-training output and iterate, rather than hold back for a finished flagship. Whether that approach closes the gap with GPT-6 and Mythos is something only independent benchmarks will settle.
