Gemini 3 and Antigravity: Why Google’s New AI Releases Are a Huge Milestone

Google has released Gemini 3, rolling out what it calls its most advanced AI model across the company’s ecosystem. The launch also includes a new programming platform called Antigravity, and for the first time, Google has integrated its latest Gemini model into search on launch day.

The moment matters. Just seven months after Gemini 2.5 and less than a week after OpenAI’s release of GPT 5.1, the timeline highlights the rapid pace at which leading AI companies are advancing their technology. Google is making Gemini 3 available immediately via the Gemini app — which has more than 650 million monthly users — and developer platforms including AI Studio and Vertex AI.

Also read: Google leader sees risk in the AI ​​race and says the sector is going through an “irrational” phase

“With Gemini 3, we are seeing a huge leap in reasoning capabilities,” said Tulsee Doshi, head of product for the Gemini model at Google. “He responds with a level of depth and nuance that we haven’t seen before.”

Gemini 3: What really changed

Let’s be nerds for a second. Here are the numbers for Google’s newest LLM:

• Gemini 3 Pro scored 1501 Elo in the LMArena rankings, beating virtually all other LLMs, including Claude, ChatGPT and Grok.

• In the GPQA Diamond benchmark, which tests scientific reasoning at the PhD level, it achieved 91.9% — better than Claude Sonnet 4.5 and ChatGPT 5.1.

• The model also scored 37.5% on Humanity’s Last Exam without tools, surpassing the GPT-5 Pro’s previous record of 31.64%.

• In math, Gemini 3 set a new standard with 23.4% in MathArena Apex.

What sets this release apart is Google’s focus on “agentic” capabilities — the model’s ability to plan and execute complex tasks with less human intervention. Demis Hassabis, CEO of Google DeepMind, described Gemini 3 as an evolution from “simply reading text and images to reading context.”

The model combines what Google calls next-generation reasoning with multimodal understanding, processing text, image, video, audio and code simultaneously.

This multimodal strength appears in tests focused on visual and video understanding. The Gemini 3 scored 81% on MMMU-Pro and 87.6% on Video-MMMU — higher scores than Claude Sonnet 4.5 and ChatGPT 5.1.

The model can also generate what Google calls a “generative user interface” — designing personalized interfaces in real time based on prompts, from interactive physics simulations to mortgage calculators.

What is Google Antigravity?

Antigravity represents Google’s effort to reinvent development tools for this new generation of AI.

Unlike previous models, where a chatbot sits in a corner of the screen waiting for questions, Antigravity puts AI in control of a dedicated workspace. AI can analyze your code, understand what you’re trying to build, write code, test, and detect problems — all with less human intervention.

Instead of treating AI as an assistant within the editor, Antigravity elevates agents to a dedicated layer with direct access to the editor, endpoint, and browser.

“With Gemini 3’s advanced reasoning, tooling, and agentic coding, Google Antigravity transforms AI from a developer tool to an active partner,” Google said in the announcement. The platform allows agents to plan and execute software tasks autonomously, while validating the code itself.

Antigravity is free during the public trial period, although users report usage limits that reset every five hours. The platform includes access to Gemini 3 Pro, Anthropic’s Claude Sonnet 4.5, and OpenAI’s GPT-OSS.

Sewing Gemini and quest

Search integration marks another important change. “This is the first time we have launched Gemini in search on day one,” wrote Sundar Pichai, CEO of Alphabet, in the announcement.

Paid Google AI Pro ($20/month) and Ultra ($250/month) subscribers can access Gemini 3 in Search’s AI Mode, which uses the model’s reasoning capabilities to generate visual layouts with interactive elements.

“Gemini 3 is also making search smarter by reimagining what a useful answer should look like,” Stein said. “With its new generative UI capabilities, Gemini 3 in AI mode can now dynamically create the entire response layout — completely in real time.”

Google is also launching Gemini 3 Deep Think, an enhanced reasoning mode that improves performance on complex problems. The mode achieved 41% on Humanity’s Last Exam without tools and 93.8% on GPQA Diamond — both category-leading results.

Deep Think will become available to Google IA Ultra plan subscribers after additional security testing.

Large development platforms are integrating the model. GitHub reported that Gemini 3 Pro demonstrated 35% more accuracy in solving software engineering challenges than Gemini 2.5 Pro in initial testing. JetBrains saw an improvement of over 50% in the number of tasks solved.

The model is also being integrated with Cursor, Manus, Replit and other programming tools. There’s a lot of optimism online right now — but, of course, it’s too early to draw definitive conclusions.

The launch seeks to return Google to a competitive position lost when ChatGPT was launched at the end of 2022.

After initial criticism of Gemini and stumbles with AI search responses, the company has hit the right path. Today, these responses reach 2 billion monthly users, and more than 70% of Google Cloud customers use the company’s AI products.