Google is accelerating development of Gemini 4 as competition among the world’s leading AI companies intensifies. The next-generation model is already in early post-training, according to Google DeepMind chief Koray Kavukcuoglu, who said the company intends to release an initial version quickly and aims to launch the model well before the end of the year.

Gemini 4 Moves Into the Post-Training Phase

Gemini 4 has moved beyond the initial training stage and into post-training, a phase in which the underlying model is refined and fine-tuned to improve performance, reliability and its ability to respond effectively to users. Kavukcuoglu said Google is already seeing promising results and wants to make an early post-training version available as soon as possible.

The approach marks a shift toward faster model development cycles. Rather than waiting for a fully optimized version before making the technology available, Google plans to release an early iteration and continue improving it rapidly. The strategy reflects the increasingly competitive pace of the AI industry, where new models are being evaluated against rivals soon after release.

Google Wants to Close the Gap With AI Rivals

The timing of Gemini 4 is significant for Alphabet. Google remains one of the largest AI infrastructure and software players, but competition from Anthropic and OpenAI has intensified as businesses increasingly evaluate models based on reasoning, coding, agentic capabilities and reliability.

Alphabet CEO Sundar Pichai previously said Google had begun what it described as its most ambitious Gemini 4 pre-training effort and that the company wanted the model to compete at the frontier level when it arrived. The latest comments indicate that development has progressed into a later stage, although Google has not provided a specific public release date.

Gemini 4 Could Become a Test of Alphabet’s AI Investment

For Alphabet, the importance of Gemini 4 extends beyond the performance of a single model. Google has been investing heavily in AI infrastructure, data centers, custom chips and product integration across Search, Cloud and other businesses. A competitive flagship model could strengthen the commercial case for those investments by supporting demand for Google Cloud services and expanding the use of Gemini across Google’s consumer and enterprise products.

The company’s recent model releases also show how quickly AI capabilities are evolving. Google has continued to introduce new Gemini Flash versions, including Gemini 3.8 Flash in September, while the development of Gemini 4 represents a move toward a larger frontier model designed to compete at the highest level.

The next major focus will be the quality of the initial Gemini 4 release rather than the timing alone. Investors and enterprise customers will be watching independent benchmarks, coding and reasoning performance, adoption across Google products and demand for Gemini-related Cloud services. A rapid release followed by frequent upgrades could allow Google to respond more quickly to competitors, but the model will ultimately be judged by measurable performance and commercial adoption. For Alphabet, Gemini 4 therefore represents both a product launch and an important test of whether its large-scale AI investment can translate into sustained competitive momentum.