Google says the new Gemini 3 AI model will understand your requests better

twin 3, Google’s latest AI model, the company’s most intelligent model yet, with more advanced multimedia and Atmosphere coding The company said in a blog post on Tuesday. It’s available now.

Google says Gemini 3 is “designed to understand depth and nuance” and is better at understanding the intent behind a user request. The company also touted the Gemini 3’s multimedia capabilities, such as its ability to turn a long video lecture into interactive flash cards or analyze someone’s pickleball game and find areas for improvement.

Gemini 3 is not limited to the app. It will also be available at Artificial intelligence mode In search, and for Pro and Ultra subscribers, in AI overviews. In AI Overviews, Gemini 3 can create interactive elements.

Don’t miss any of our unbiased technical content and lab reviews. Add CNET As Google’s preferred source.

The Gemini 3 Pro is less flatter, an issue that plagues AI systems and leads to AI psychosis in some, Google DeepMind’s Demis Hassabis and Koray Cavukoglu said in a blog post. It is also safer against Rapid injection attackswhere bad actors attempt to make the AI ​​ignore its original instructions and perform unintended actions.

The company also unveiled Google Antigravity, a new platform for agent development. Google says Antigravity is like having an active partner while making tools or working on projects, independently planning and executing complex software tasks while validating their code. It works alongside and works with Gemini 2.5’s computer usage model to control the browser Nano bananaGemini Image Model 2.5.

The new, more powerful proxy capabilities in Gemini 3 will be available for just $250 per month Google AI Ultra Subscribers at the beginning. This will allow the Gemini agent to do a multi-step workflow, such as planning Itinerary.

Read more: AI basics: 29 ways you can make AGI work for you, according to our experts

Google’s release of Gemini 3 comes as the AI ​​war between it and OpenAI, Anthropic, and xAI heats up. Google has always been a leader on AI leaderboards, although other AI models haven’t been far behind, sometimes trading spots at the top. With the release of Gemini 3, Google appears to be trying to solve some of the most annoying AI problems, like hallucinations or ingratiation. It also attempts to prove that AI can be truly effective, being able to accomplish tasks on behalf of the user. Other proxy models have proven and are problematic to use in the real world Various security concernsEspecially in web browsers.

Google’s latest AI release also comes at a time when there are fears of an AI bubble forming in the stock market. It represents AI companies, including Nvidia, Google, Meta, and Microsoft 30% of the Standard & Poor’s 500 Index. Google currently has a rating $3.4 trillion. Even Google CEO Sundar Pichai says the trillion-dollar AI investment boom has “irrational elements” and that the boom will impact every company in the AI ​​space, he said in an interview with The Sun. BBC. However, if Google wants to keep its stock price high, it needs to prove that its AI models outperform the competition.