Google has officially launched its latest and most advanced AI model, Gemini 3 – and the company is calling it its “most intelligent” system yet. The model launched on Google Products on the very first day, marking the first time that a new Gemini release has gone live in Search at launch.
Gemini 3 represents what Google executives describe as their next major step toward more capable, more context-aware AI. In a note sharing the launch, the company said the model includes deeper reasoning, stronger multimodal understanding, and increased awareness of user intent compared to earlier versions.
What’s new in Gemini 3?
The title upgrade is logical. Google says Gemini 3 is built to understand nuance, solve complex problems, and “understand depth and context” in a way its predecessor couldn’t. In benchmarking tests, the Gemini 3 Pro outperformed both the Grok and the Gemini 2.5 Pro, topping the LMArena leaderboard with a 1501 Elo score.
Google’s Gemini 3 is smarter, better at figuring out what you need
The model also represents a significant improvement over academic-style reasoning. On Humanity’s last test, the Gemini 3 Pro achieved 37.5 percent without any external devices, and on GPQA Diamond, it reached 91.9 percent. It also posted state-of-the-art scores in mathematics and multimodal tasks, including video comprehension and factual accuracy.
mashable light speed
A new advanced mode called Gemini 3 Deep Think extends those capabilities even further. Deep Think improves scores on nearly every benchmark, including a 41 percent increase on the Ultimate Exam in Humanity and a 45.1 percent increase on the ARC-AGI-2 challenge. It is designed for the most challenging, long-term reasoning tasks.
Where can you use Gemini 3?
Gemini 3 is already live in the Google ecosystem. These include:
-
AI mode in searchWith more dynamic, generative interfaces and complex logic support. (Currently available to Google AI Pro and Ultra customers.)
-
gemini app The My Stuff folder is getting a new design, including better shopping experiences.
-
gemini for developersNow available in AI Studio, Vertex AI, Gemini CLI, and Google’s new agentic development platform AntiGravity.
-
gemini agentAn advanced system that can accomplish multi-step tasks like inbox organization, appointment management, and more.
How to try Gemini 3 now
If you want immediate hands-on time with Gemini 3, there are a few ways to go:
-
Use AI mode in search (Google AI Pro or Ultra subscription required).
-
Open Gemini AppWhere Gemini 3 is the default model for most users globally starting today.
-
developers Can experiment with AI Studio or Vertex AI, while gaining full access to Gemini 3 Pro.
-
enterprise customer It can be deployed through Vertex AI and Gemini Enterprise.
Deep Think, the highest-level version, will be launched in the coming weeks after additional security evaluation.