The short version
- Koray Kavukcuoglu states that Gemini 4 is in its final refinement phase and will launch significantly before the end of the year.
- Google has not released a flagship model since November 2025, allowing competitors like OpenAI and Anthropic to establish a lead with newer systems.
- The delay followed a strategic pivot toward less powerful Flash models and a leadership change after Demis Hassabis stepped down in August.
Google is preparing to release its next-generation artificial intelligence model, Gemini 4, much sooner than previously anticipated. Koray Kavukcuoglu, who recently assumed leadership of Google’s DeepMind division, indicated during his first media appearance that the model is currently undergoing final refinement. He expressed confidence in the results seen so far and stated that the company intends to release an early post-training output as soon as possible. This announcement marks a shift in pace for the tech giant, which has faced criticism for lagging behind competitors in recent flagship AI releases.
The timeline for Gemini 4 represents a significant acceleration compared to earlier expectations. Kavukcuoglu told The Information that the launch is expected to occur well before the end of the year. He emphasized that Google plans to maintain a rapid iteration cycle moving forward, suggesting a renewed commitment to consistent development schedules. This approach aims to address concerns about the company’s ability to keep pace with the fast-moving landscape of generative AI technology.
Google’s absence from the forefront of flagship model releases has been notable over the past several months. The company last introduced a major update with the Gemini 3 series in November 2025. Since that time, rival firms have advanced their offerings considerably. OpenAI launched its GPT-6 model, and Anthropic released its Mythos series. Both of these newer systems have demonstrated performance levels that surpass those of Gemini 3, granting their developers a substantial advantage in the competitive race for AI supremacy.
The delay in releasing new flagship models stems partly from a strategic decision to prioritize different product lines. Earlier this year, Google CEO Sundar Pichai announced at the I/O event in May that a significant update, Gemini 3.5 Pro, would arrive in June to help bridge the performance gap with competitors. However, that model was never rolled out. Instead, Kavukcuoglu explained that the company took a step back from pursuing the high-end update to focus on introducing faster, less powerful Flash models. This pivot allowed Google to optimize for speed and efficiency rather than raw capability during this interim period.
Leadership changes within DeepMind have also played a role in the current trajectory of AI development at Google. Demis Hassabis, the former head of AI research, stepped down from his position in August. At the time of his departure, he noted that great progress was being made on Gemini 4. The transition to Kavukcuoglu appears to have coincided with a renewed focus on bringing this next-generation model to market. His public comments suggest an intent to stabilize release timelines and restore confidence in Google’s AI roadmap following the recent shakeup at the top.
Despite the competitive pressure, Kavukcuoglu remains confident in Google’s long-term position in the industry. When asked about concerns that the company might be falling behind rivals in AI development, he dismissed such worries. He stated that it is a certainty in his mind that Google will always remain at the frontier of technological innovation. This assurance reflects a broader corporate belief that the underlying research and infrastructure are sufficient to catch up with and eventually surpass current market leaders.
The upcoming release of Gemini 4 will be closely watched by industry analysts and competitors alike. Its performance characteristics, particularly in comparison to GPT-6 and the Mythos series, will determine whether Google can reclaim its status as a primary driver of AI advancement. The model’s capabilities in reasoning, coding, and multimodal processing are expected to be key differentiators. If successful, the launch could signal the end of a period of strategic hesitation and the beginning of a more aggressive phase in Google’s AI competition.
Looking ahead, the focus will shift from development timelines to real-world application and user adoption. The success of Gemini 4 will depend not only on its technical benchmarks but also on how effectively it integrates into existing Google services and third-party platforms. As the company continues with fast-paced iterations, stakeholders will monitor whether this renewed momentum translates into sustained market leadership or merely a temporary catch-up effort. The coming months will be critical in defining the next chapter of the global AI race.
Sources behind this briefing
Go to the original reporting
- The Verge↗Gemini 4 is almost ready, says new Google DeepMind chief
