Google officially announced Gemini 3, its new big move that intensifies the competition in the field of artificial intelligence. Unlike previous versions, the company integrated its new model directly into its own search engine and developer infrastructure on the day it introduced it for the first time. Thus, Gemini 3 was established in the Google ecosystem from the moment of launch.
“The smartest artificial intelligence ever”
While Google CEO Sundar Pichai described the model as “the smartest artificial intelligence ever”, he stated that Gemini 3 is no longer just a system that produces content, but offers a significant leap in understanding context and user intent.
93.8 percent score in the GPQA Diamond test
Technical developments also support this claim. According to the internal test results shared by Google, Gemini 3’s Deep Think mode achieved significant success in difficult evaluations requiring reasoning. Model, 41 percent on Humanity’s Last Exam; In the GPQA Diamond test, it scored 93.8 percent, leaving Anthropic and OpenAI’s current most advanced models behind.
At the event, Google introduced not only Gemini 3 but also the long-developed Gemini Agent. Google CTO Koray Kavukçuoğlu said that this new tool can complete multi-step processes, from organizing emails to travel planning, without the need for user intervention.
The company also announced new interface updates for developers. While the new platform named “Antigravity” was opened for preview, it was stated that this system allows artificial intelligence agents to make comprehensive planning for projects and produce the necessary code independently.
Gemini application has reached 650 million users.
The effects of Google’s strategy of placing artificial intelligence at the center of the ecosystem are also reflected in the figures. According to the data shared by the company, the AI Overviews feature has reached 2 billion monthly users, while the Gemini application has reached 650 million users.
