Google has a new contender for the smartest AI in the world.
On Tuesday, the company launched the Gemini 3 and described it as its “most intelligent model.”
The new LLM (large language model) should have improved reasoning skills, which will help it understand the depth and nuances of the tasks it deals with. But according to Google, it should also be better at detecting “the context and intent behind your request,” which means you should be able to make do with fewer signals.
Google Maps introduces conversational navigation with Gemini
Gemini isn’t just updated on its own. Google said it has shipped LLM extensively across all of its products, meaning Gemini 3 is already live in AI mode in Google Search — the first time the company has shipped a new model in Search on launch day (note, though, it’s currently only live for Google AI Pro and Ultra customers). Gemini 3 is now live in Gemini Apps, AI Studio, and Vertex AI, as well as Google’s agentive development platform, AntiGravity.
mashable light speed
As is customary, Google immediately showed the Gemini 3’s scores in various benchmarks. For example, the Gemini 3 Pro now leads in LMArena with a score of 1501 Elo, ahead of XAI’s Grok and the Gemini 3’s predecessor, the Gemini 2.5 Pro.
And if you want to take it a little further, you can run Gemini 3 DeepThink, which is even better at reasoning and more suitable for the most complex tasks. That model is significantly better in some benchmarks, including the notoriously difficult Ultimate Test of Humanity.
Apart from all this, Google also redesigned its Gemini app. One major change is the My Stuff folder, where you can find various chats and documents created with Gemini. The company also improved the shopping experience, and began experimenting with generative interfaces, which are created by models based on what you prompt.
Finally, Gemini Agent is a feature that can complete multi-step tasks for you, including managing your appointments and reminders, organizing your inbox, or conducting online research.