Google prepares to unveil Gemini 4 AI model

Google is preparing to release its long-awaited Gemini 4 artificial intelligence model. Koray Kavukcuoglu, the new head of Google DeepMind, said the model is currently being refined. The company plans to make Gemini 4 available well before the end of 2026.

Speaking to the media for the first time in his new role, Kavukcuoglu said the model's results have generated considerable interest within the company. Google wants to release the initial post-training version as soon as possible. The model will then be updated frequently, with its capabilities expanded in stages.

Google introduced its latest major AI model, the Gemini 3 series, in November 2025. OpenAI's GPT-6 and Anthropic's Mythos models have been released since then. Both models have outperformed Gemini 3 and strengthened the rival companies' positions in the AI race.

Google CEO Sundar Pichai announced at the I/O event in May that the Gemini 3.5 Pro update would be released in June. However, the model never reached users. According to Kavukcuoglu, instead of releasing the flagship model, the company prioritized faster but relatively weaker Flash models.

The leadership change at Google DeepMind took place in August. The division's former head, Demis Hassabis, stepped down after announcing significant progress on Gemini 4. Kavukcuoglu said he is confident that Google will remain among the leading companies in artificial intelligence.