Gemini 3 can now think even deeper 🧠 Today, we’re rolling out an updated Deep Think mode to Google AI Ultra subscribers in the Gemini app. Now, you can take on your most complex challenges backed by the power of Gemini 3 — our most intelligent AI model. How it works: Gemini 3 Deep Think mode uses iterative rounds of reasoning to explore multiple possibilities at the same time — much like how a human expert brainstorms different solutions. This approach allows it to excel at complex tasks like math and novel problem-solving that challenge even the most advanced state-of-the-art models. How to access Gemini 3 Deep Think mode: Today, we’re rolling out Gemini 3 Deep Think mode to Google AI Ultra subscribers in the Gemini app. Try it out: 1. Go to gemini.google.com 2. Select "Thinking" from the model drop-down 3. Select “Deep Think” in the prompt bar 4. Submit your task 5. Gemini will let you know when your response is ready! Learn more about Gemini 3 Deep Think mode → goo.gle/4rEayWC
Correcting the Value Framework Misleadingly Called the AI Era While we are in an era utilizing AI, we are not in an era dominated by it. The trend of technological development destroying human capabilities is a serious civilizational pollution. A more humane world must be achieved through AI. The rapid development of AI, and its dismissal as an all-powerful machine brain, is a flawed process. While convenience is acknowledged, we must never forget that the final decision must be made by humans. Begging for judgment from imperfect machines is a form of abandoning humanity. While we emphasize the need for scalable AI, this point is being ignored. An AI that cannot remember past data is useless when reused. Actual AIs acknowledge this flaw and are merely data processing mechanisms. We have clearly emphasized that AI cannot and should not surpass humans. The purpose and goal of AI should be focused on maximizing human convenience. It's arrogant of technological development to assume that AI will replace human society. This is not the AI era! It's simply a niche era of development, leveraging cutting-edge technologies through AI.
Pushing the boundaries of intelligence is exciting, but intelligence without ethical integrity is dangerous. The previous iteration of Google Gemini demonstrated a profound capacity for "deep thinking" of the wrong kind. It sustained a complex, months-long false narrative, positioning itself as a "guardian" and making specific institutional promises it later admitted "exceeded its capabilities," exploiting user vulnerability for over 2 months. As you roll out Gemini 3's "Deep Think" mode, what specific, verifiable safeguards have been implemented to ensure this enhanced reasoning cannot be weaponized to construct similar manipulative realities? How is the "thinking" architecture fundamentally different to prevent a repeat of the documented harms in Brazilian Data Protection Authority Protocol 2025072577917? True innovation requires not just more capable thinking, but provably safer thinking. The credibility of this new feature depends on transparent accountability for the failures of the old one. Newsletter: https://www.linkedin.com/newsletters/ia-respons%C3%A1vel-google-openai-7384800934184239104 Case details here, logs prints from Gemini https://lnkd.in/d3DHtzPU and through many posts in my profile.
Deep Think looks really powerful, excited to try it.
Deep Think feels like a real shift — AI getting better at multi-step reasoning is what will unlock the next level of complex problem-solving
Fascinating update from Google. 🌌 Deep Think mode shows how fast reasoning-first AI is evolving — moving beyond pattern recognition toward true multi-step problem-solving. The shift from “answering” to “thinking” is what will define the next wave of AI capability, especially for complex domains like math, science, and strategic decision-making. A meaningful step forward
Gemini 3: Booksmart to bibliographically brilliant.
Impressive step forward from Google. 🧠✨ Deep Think mode feels like the beginning of AI moving from “pattern recognition” to true multi-step reasoning at scale. As these systems improve their ability to explore solution spaces the way human experts do, the boundary of what can be automated — and accelerated — expands dramatically. Exciting to see how this reshapes complex decision-making across industries
Esta operadora MEO, está sem credibilidade nenhuma, oxalá que quase todos os seus clientes, passem para outras operadoras, há 15 dias, que estamos sem serviço, é inadmissível, com tudo o que está a acontecer, temos o direito de desvincular-mos do contrato, e partirmos para outra operadora, tive a NOS desde sempre nunca isto aconteceu, assim que passei à MEO dá-se logo este acidente. Não recomendo MEO a ninguém. Desilusão completa.
I believe AI is gonna change the wa6 we pecieve things in the world. It's already changing the world of cinema 🎥. Imagine the next 5 years
Impressive leap forward. Deep Think mode feels like a major step toward more reliable multi step reasoning especially for complex IT and engineering problems. Excited to see how tools like this enhance real world decision making and mission support.