Google DeepMind

Hello from Google DeepMind’s newest AI research lab in Singapore! Delighted to announce this milestone with our growing team & regional partners, including Singapore Economic Development Board (EDB) and AI Singapore. Singapore's pioneering outlook thoughtfully balances long- and short-term opportunities & risks—alongside a unique, diverse, and innovative culture and its regional and global leadership role. Our new lab builds on collaborative efforts such as the SEA-LION initiative and Project Aquarium with AI Singapore to advance foundational research on linguistic and cultural inclusivity in our models and products. We're also accelerating our real-world impact across AI for science and learning in close partnership with local ecosystems, building on Google's longstanding commitment to the region. Having spent 7 years (across 3 decades) living in Asia, I feel deeply connected to the region. Looking forward to deepening our connections & working together to translate AI breakthroughs into tangible, inclusive benefits for everyone! 

Starting today, you can ask the Gemini app: "Is this image made with AI?" 💬 Upload a picture and it will check for the SynthID watermark to help verify if it was generated or edited by Google AI tools. See the details → https://goo.gle/49vZxjC

We just dropped Nano Banana Pro, built on Gemini 3. 🍌 With enhanced reasoning and real-world knowledge, Gemini 3 Pro Image can help you create detailed diagrams, infographics and more. It can also: 🔹 Generate clear and accurate text in multiple languages. 🔹 Modify lighting, camera angles, allowing you to change focus, colors and more. 🔹 Export in a range of aspect ratios, with outputs in 1K, 2K. and 4K. 🔹 Combine multiple photos to create anything from a product mock-up, to a storyboard, all from one prompt. Nano Banana Pro is the ultimate partner for building out your creative vision. Start trying it in Gemini App, Google AI Studio, AI Mode in Google Search, and more → https://goo.gle/43Hh1Wu

We used Gemini 3 to create bite-sized demo games on YouTube Playables with just a few simple prompts 🎮 Give them a try → https://goo.gle/485X5Ow

Gemini 3 is here, and it's built to be our most secure model yet. 🔒 As we build our frontier AI models, we’re constantly thinking about how to ensure they benefit humanity and minimize potential harms. This model has undergone the most comprehensive set of safety evaluations of any Google AI model. We’ve also focused on three key areas: ✅ Rigorous testing against our Frontier Safety Framework ✅ Independent assessments with world-leading subject matter experts ✅ Increased resistance to prompt injections and robust protection against cyberattacks See our Gemini 3 model card → https://goo.gle/4r8ltrl

Google Antigravity is our new agentic development platform. AI is your active partner to help you build faster, with agents able to autonomously plan and execute tasks across the editor, terminal, and browser. It uses Gemini 3 Pro to reason about problems, Gemini 2.5 Computer Use for end-to-end execution, and Nano Banana for image generation. Developers can download the public preview at no cost now → https://lnkd.in/ewphHXmc

Bring any idea to life with Gemini 3: our most intelligent model, designed to help you learn, build and plan anything. We’re first releasing Gemini 3 Pro, which is rolling out globally starting today. This is how we’re pushing the frontier: 🔵 State-of-the-art reasoning: It understands prompts with incredible depth and nuance, delivering clear, direct answers without clichés or filler. As our most factual model, it’s more reliable for complex questions in science and math. 🔵 World-leading multimodal understanding: Gemini 3 seamlessly comprehends text, images, video, audio, and code. It adapts to you, responding with whatever best suits your needs. Quickly turn text lessons into visual flashcards or ask Gemini to break down concepts from a long video. 🔵 Our best model for vibe and agentic coding: You can build dynamic, beautiful apps from a single prompt. We’ve also improved agentic code performance – supporting existing tools and our new agentic development platform, Google Antigravity. We can’t wait to see what you build. Here’s how you can try it in Gemini app, Google AI Studio, AI Mode in Search, and Google Cloud’s Vertex AI for enterprises → https://goo.gle/Gemini-3

Weather affects everything and everyone. Our latest AI model developed with Google Research is helping us better predict it. ☔ WeatherNext 2 is the latest in our family of weather models, able to generate more accurate and higher-resolution global forecasts. One core challenge in this field is capturing the full range of possible outcomes. With WeatherNext 2, we can: ☁️ Generate a full range of possible forecast scenarios in a single step. ☁️ Deliver up to 1-hour resolution global forecasts faster than ever. ☁️ Improve the efficiency of forecasting with an 8x increase in speed. ☁️Show superior skill across 99.9% of weather variables such as: temperature, wind, humidity, and pressure levels, compared to our previous model. WeatherNext 2 marks a significant leap for the scientific community and beyond. We’ve now incorporated WeatherNext technology on Google Search, the Gemini app, Pixel Weather and more - and in the coming weeks, it will also help power weather information in Google Maps. Find out more → https://goo.gle/3WWdKii 

We’re introducing SIMA 2, the next major milestone in general and helpful embodied AI agents. 👾 With Gemini integrated at its core, it moves beyond following basic instructions to think, learn, and collaborate in complex, 3D worlds. 🔵 Advanced reasoning: It can accomplish high-level goals in a wide array of games – describing its intentions, explaining what it sees, and outlining the steps its taking. 🔵 Improved generalization: It can transfer concepts like “mining” in one game and apply it to “harvesting” in another - connecting the dots between similar tasks. 🔵 Self-improvement: Through trial-and-error and Gemini-based feedback, it can teach itself entirely new skills in unseen worlds without additional human input. 🔵 Adaptability: When tested in simulated 3D worlds created with our Genie 3 world model, it demonstrates unprecedented adaptability by navigating its surroundings, following instructions, and taking meaningful steps towards goals.  This research offers a strong path toward applications in robotics and another step towards AGI in the physical world. Learn more → https://goo.gle/SIMA-2