Google CEO Sundar Pichai confirmed at the latest earnings call that the company plans to launch its next-generation AI model, Gemini3, this year. The new model aims to surpass the current Gemini2.5Pro and focus on enhancing its **"Agent"** capabilities in handling complex, multimodal tasks, in an effort to further close the performance gap with top competitors like OpenAI's GPT-5.

Gemini3: Focused on Complex Multimodal Tasks

Pichai pointed out that Gemini3 will place greater emphasis on practicality and complexity in its design. It is not only expected to outperform existing models in various performance metrics, but it will also make processing multimodal data (such as text, images, and code) and executing complex agent tasks its core strengths. This means that the future Gemini3 will be better able to understand and handle users' coherent, in-depth cross-domain needs.

Although Pichai praised the team's execution capability and the speed of model progress, he also admitted that further breakthroughs in cutting-edge AI models require a longer development cycle. He reminded the market that Gemini3's progress will not be a "overnight" leap, indicating that the company maintains a rational understanding of the laws of technological development while pursuing performance excellence.

Google's Large Model Gemini

AI Drives Performance Growth: Quarterly Revenue Surpasses $100 Billion for the First Time

Notably, Google's parent company Alphabet released its strongest financial report ever last week. In the third quarter of 2025, Alphabet successfully achieved a milestone of $100 billion in single-quarter revenue, which exceeded market expectations across the board.

The report emphasized that artificial intelligence has become the core driver of substantial growth across all of Alphabet's businesses. This strategic investment has brought significant commercial results:

  • Strong user growth: The monthly active users (MAU) of Google's AI application Gemini App have surpassed 650 million, and query volume in the third quarter increased by three times compared to the previous quarter.

  • Deepened ecosystem collaboration: The renowned AI research company Anthropic announced plans to use up to 1 million of Google's TPUs (Tensor Processing Units) for model training, demonstrating the strong appeal and market leadership of Google's AI infrastructure.

This series of positive data not only solidifies Alphabet's position among global tech giants, but also provides a solid financial foundation for its continued high investment in AI.