{"id":34103,"date":"2025-11-19T08:47:54","date_gmt":"2025-11-19T08:47:54","guid":{"rendered":"https:\/\/metaverseplanet.net\/blog\/?p=34103"},"modified":"2026-01-05T08:01:05","modified_gmt":"2026-01-05T08:01:05","slug":"google-unveils-gemini-3","status":"publish","type":"post","link":"https:\/\/metaverseplanet.net\/blog\/google-unveils-gemini-3\/","title":{"rendered":"Google Unveils Gemini 3, Its Most Advanced AI Model"},"content":{"rendered":"\n

Google has introduced Gemini 3<\/strong>, which stands out with its multimodal intelligence<\/strong>, advanced reasoning<\/strong>, and coding capabilities<\/strong>. While the model is available for use starting today, the Deep Think mode<\/strong> will be available soon.<\/p>\n\n\n\n

A New Standard in AI Intelligence<\/strong><\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n

Starting today, Google has begun rolling out its new AI models, named Gemini<\/a><\/em> 3<\/strong>, to users. The company defines this series as the “smartest” AI systems with high factual accuracy<\/strong>. Gemini 3 is seen as a crucial step for Google to take the lead in consumer-focused AI<\/strong>, especially following the issues with OpenAI<\/strong>‘s GPT-5<\/strong> launch. The DeepMind<\/strong> team, led by Demis Hassabis<\/strong> and Koray Kavuk\u00e7uo\u011flu<\/strong>, describes Gemini 3 as “our smartest model capable of bringing every idea to life.”<\/p>\n\n\n\n

Introduced just seven months after Gemini 2.5<\/strong>, this new version stands out as Google’s most capable large language model (LLM)<\/strong> to date. The most notable member of the new series, Gemini 3 Pro<\/strong>, has been opened to all users via the Gemini app<\/strong> starting from the launch day. Additionally, Google AI Pro<\/strong> and Ultra subscribers<\/strong> in the US will be able to experience Gemini 3 Pro via AI Mode<\/strong> within Google Search<\/strong>.<\/p>\n\n\n\n

\n\n\n\n

Multimodal Capabilities and Product Integration<\/strong><\/h2>\n\n\n\n

Gemini 3 Pro<\/strong> is designed as a naturally multimodal model<\/strong>. This means it can process text<\/strong>, image<\/strong>, and audio<\/strong> data simultaneously. For example, it can take photos of a recipe and turn them into a cookbook or create interactive cards from a video lecture series.<\/p>\n\n\n\n

The improvements of the new model will be felt across most Google products<\/strong>. In the Gemini app<\/strong>, users gain a workspace where they can create more comprehensive programs. Additionally, thanks to generative interfaces<\/strong> tested in Gemini Labs<\/strong>, Gemini 3 Pro can design visual and magazine-style content or custom user interfaces. In Search with AI Mode<\/strong>, responses to your queries are presented with visual elements<\/strong> such as images, tables, grids, and simulations.<\/p>\n\n\n\n

\n\n\n\n

Smarter, More Direct, and Reliable<\/strong><\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n

Google<\/a><\/em><\/strong> emphasizes that thanks to the updated “query chunking<\/strong>” technique, the model can better understand the intent of questions and find content missed in previous searches. Google defines Gemini 3 Pro as a more reliable model compared to ChatGPT<\/a><\/em><\/strong>‘s empty praises. Google DeepMind VP of Product Tulsee Doshi<\/strong> states that answers are now smart, concise, and direct, offering real information<\/strong> instead of clich\u00e9s and excessive compliments. In other words, the model now engages in much less sycophancy<\/strong>.<\/p>\n\n\n\n

Gemini 3 Pro\u2019s reasoning<\/strong> and task completion<\/strong> capabilities have also been improved. It can perform more complex tasks and engage in long-term planning<\/strong>. These features are used in the Gemini Agent<\/strong> experimental feature, which can perform operations such as email review, travel research, and booking on behalf of the user.<\/p>\n\n\n\n

\n\n\n\n

Advanced Logic: “Deep Think” Mode<\/strong><\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n

Furthermore, the Deep Think mode<\/strong>, which is currently open only to test users, further increases reasoning capacity<\/strong>. After passing safety tests, Deepthink will be offered to Google AI Ultra<\/strong> subscribers in the coming weeks.<\/p>\n\n\n\n

Deep Think mode<\/strong> takes the model’s logic and multimodal understanding capacity even further. In tests, Gemini 3 Deep Think<\/strong> surpasses Gemini 3 Pro with scores of 41% in Humanity\u2019s Last Exam<\/strong> and 93.8% in GPQA Diamond<\/strong>. It also demonstrates its ability to solve new and unique challenges with 45.1% in ARC-AGI-2<\/strong>.<\/p>\n\n\n\n

\n\n\n\n

Powerful Tools for Developers: Antigravity<\/strong><\/h2>\n\n\n\n

Gemini 3 offers powerful tools for developers<\/strong>. With its zero-shot production capability, it can process complex instructions to create richer and more interactive web interfaces<\/strong>. It achieves an Elo score of 1487 in WebDev Arena<\/strong>, making it the most competent model in agentic coding<\/strong> and vibe coding<\/strong>. It provides a significant advantage over 2.5 Pro in Terminal-Bench 2.0<\/strong> and SWE-bench Verified<\/strong> tests.<\/p>\n\n\n\n

Introduced simultaneously with Gemini 3, the coding interface named Antigravity<\/strong> offers a multi-panel<\/strong> and agent-assisted<\/strong> experience. Working similarly to tools like Warp<\/strong> or Cursor 2.0<\/strong>, Antigravity combines a ChatGPT<\/strong>-style command window, a terminal interface, and a browser window showing the effects of changes. The model is also accessible on third-party platforms such as Cursor<\/strong>, GitHub<\/strong>, JetBrains<\/strong>, Manus<\/strong>, and Replit<\/strong>.<\/p>\n\n\n\n

\n\n\n\n

Breaking Records in Benchmarks<\/strong><\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n

Looking at the performance side, the model also shows strong performance in independent measurements. It achieved the highest score of all time with 37.4 points in Humanity\u2019s Last Exam<\/strong>. It also ranks at the top in LMArena<\/strong>, which measures user satisfaction. It set a new standard with 23.4% in the field of mathematics at MathArena Apex<\/strong>. Standing out with its multimodal skills<\/strong>, Gemini 3 Pro offers reliable solutions to complex problems with 81% success in MMMU-Pro<\/strong>, 87.6% in Video-MMMU<\/strong>, and 72.1% in SimpleQA Verified<\/strong>.<\/p>\n\n\n\n

\n\n\n\n

Global Availability and The Gemini 3 Era<\/strong><\/h2>\n\n\n\n
\"\"<\/figure>\n\n\n\n

According to Google, the Gemini app<\/strong> has 650 million monthly active users<\/strong>, and 13 million software developers<\/strong> are actively using the model in their workflows. The distribution of Gemini 3 has spread to a wide ecosystem as of today. While the model is available to all users in the Gemini app, Google AI Pro<\/strong> and Ultra subscribers<\/strong> can also benefit from the new capabilities via AI Mode<\/strong>. Developers<\/strong> can access Gemini 3 through the Gemini API<\/strong>, AI Studio<\/strong>, Google Antigravity<\/strong>, and Gemini CLI<\/strong>. On the corporate side, the model has been integrated into Vertex AI<\/strong> and Gemini Enterprise<\/strong> platforms.<\/p>\n\n\n\n

You Might Also Like;<\/h3>\n\n\n