Given how solid and remarkable Gemini 2.5 has been so far, expectations for Gemini 3 are very high. Well, Google is releasing its new AI model today and it seems that Gemini 3 is really smart and good enough to beat the current state-of-the-art AI models like ChatGPT 5 and Grok 4. Announcing Gemini 3, the company describes its latest AI as its most advanced system to date, capable of interpreting information with a level of depth and nuance that is close to human reasoning.

Google says it is rolling out Gemini 3 to users across all of its platforms, including Search. It’s also going to be available to all users in the Gemini app, although there will be different usage limits on it depending on which Google AI plan you’re subscribed to. Google says that more than 650 million users are currently using Gemini AI every month.

According to CEO Sundar Pichai, Gemini 3 brings together Google’s work in multimodal understanding, long context analysis, and agentic behavior into one AI system. “Gemini 3 is the cutting edge of reasoning, built for understanding depth and nuance – whether it’s picking up subtle clues in a creative idea, or separating the overlapping layers of a difficult problem,” the Google CEO said in a blog. “Gemini 3 is also much better at understanding the context and intent behind your request, so you get what you need with less prompting. It’s amazing to think that in just two years, Al has evolved from just reading text and images to reading the room.”

Google says Gemini 3 is built for logic

Google DeepMind CEO Demis Hassabis told users about the AI ​​system in another note. “Today we’re taking another big step forward in AGI with the release of Gemini 3,” Hassabis wrote. “This means Gemini 3 Pro is highly capable of solving complex problems across a wide range of subjects such as science and maths with a high level of reliability. Gemini 3 Pro also brings a new level of depth and nuance to every conversation. Its responses are smart, concise and direct, trading clichés and flattery for genuine insight.”

Google claims the Gemini 3 performs strongly on academic and synthetic tests that measure analytical depth, including the Humanities Last Exam, GPQA Diamond, and a new high score on the MathArena Apex, which examines mathematical and symbolic problem-solving.

Abilities of Gemini 3

Google says the new AI model can analyze and understand different types of content. This includes handwritten notes, long research papers, and even video lectures. According to the company, users can apply these capabilities to scenarios like converting family recipes into digital formats, creating structured notes from dense educational content, or breaking down long, complex videos into step-by-step explanations.

The model will also underpin new features in AI mode on Google Search, where it can generate dynamic visual layouts, simulations, and interactive tool-based results.

Gemini 3 Deep Think experiment

Along with the main model, Google also previewed Gemini 3 Deep Think, an extended-reasoning mode designed for demanding analytical tasks. In internal testing, Google says it performs slightly better than the Gemini 3 Pro on higher-difficulty benchmarks and shows stronger tool-use capabilities, especially when running code or validating hypotheses.

However, Deep Think Mode will not be available to users right now. Instead, Google says that this mode will be tested by researchers and select testers for now. Once the company is satisfied that Think Mode can be released safely, it will likely be rolled out to Gemini Ultra customers.

Gemini 3 brings antigravity to developers

Based on the new AI system, Google is also expanding Gemini’s agentive and coding features specifically for developers.

For developers, Gemini 3 clearly comes with stronger coding performance and more consistent agentic behavior, with the ability to plan actions for models, use tools, and take multi-step actions over a longer period of time. According to Google, Gemini 3 currently leads the WebDev Arena leaderboard and shows better results on coding-agent tests such as SWE-Bench Verified and Terminal-Bench 2.0.

Google is also launching Google AntiGravity, a new software development tool where AI agents can work more autonomously across an editor, terminal, and browser. The company says that instead of merely acting as a suggestion tool, the AI ​​agent can plan and execute the entire workflow almost like a junior developer, while the user remains in control.