Google only has a few short weeks. Naturally, the company claims the new system is its most intelligent AI model yet, offering cutting-edge logic, class-leading Vibe coding performance, and more. The good news is that you can test those claims today, Google is making the Gemini 3 Pro available in many of its products and services.
Google is highlighting some benchmarks to improve the performance of the Gemini 3 Pro. Widely considered one of the toughest tests AI labs perform on their systems, the model delivered a new top accuracy score of 37.5 percent, beating previous leader Grok 4 by an impressive 12.1 percentage points. Notably, it achieved its score without resorting to tools such as web search. Meanwhile, the Gemini 3 Pro now tops the site’s leaderboard with a score of 1,501 points.
OK, but what about the practical benefits of the Gemini 3 Pro? In the U.S., the new model will translate answers that are more concise and better formatted. It also enables a new feature that Google calls Gemini Agent. The tool is based on the web-surfing Chrome AI, which the company launched late last year. It allows users to ask Gemini to complete tasks for them. For example, let’s say you want help managing your email inbox. In the past, Gemini might have given some general suggestions. Now, this can do that job for you.
To try out Gemini 3 Pro inside the Gemini app, select “Thinking” from the model picker. The new model is available to everyone, although AI Plus, Pro and Ultra customers can use it more often before reaching their rate limit. To get the most out of Gemini Agent, you need to give the tool access to your Google apps.
Meanwhile, Search will see the debut of Gemini 3 Pro, with the new model becoming available first to AI Pro and Ultra subscribers. Google will also bring this model to AI Overviews, where it will be used to answer the most difficult questions people ask its search engine. In the coming weeks, Google plans to introduce a new routing algorithm for both AI Mode and AI Overview, which will let it know when to ask questions through the Gemini 3 Pro. Meanwhile, subscribers can try out the new model inside AI mode by selecting “Thinking” from the dropdown menu.
In practice, Google says the AI mode on the Gemini 3 Pro will result in finding more reliable and relevant content related to your queries. This is thanks to how the new model enhances the fan-out technology that powers the AI mode. The tool will search even further than before and with its new intelligence, Google suggests it can even uncover content that previous models may have missed. Additionally, Gemini 3’s improved multi-modal understanding will translate into AI mode, generating a more dynamic and interactive interface to answer your questions. For example, if you’re researching mortgage loans, the tool can build a loan calculator right inside its response.
For developers and its enterprise customers, Google is bringing Gemini 3 to all the usual places where one can find its models, including the Gemini API, AI Studio, and Vertex AI. The company is also releasing a new agentic coding app called AntiGravity. It can program autonomously while creating tasks for itself and providing progress reports. Along with the Gemini 3 Pro, Google is introducing Gemini 3 DeepThink. Advanced Reasoning Mode will be available to security testers before it is rolled out to AI Ultra customers.