Google simply skips every competitor with a stunning AI who can think deeper, shop more intelligently, and create videos with dialogue

Google Announces an extensive set of artificial intelligence progress on Tuesday at his annual annual Conference on Developers of I/OIntroducing helpful AI models, expanding your search capabilities and launching new creative tools that are running the boundaries of what its technology can achieve.

The Mountain View -based company has presented Twin improvements 2.5rolled AI mode in search To all of us users entered New generative media modelsand released premium $ 249.99 a monthly subscription level called Google have ultra For power users, all reflect the acceleration of Google AI in its product ecosystem.

“More intelligence is accessible to everyone, everywhere. And the world reacts, accepting AI faster than ever,” says Sundar Pichai, CEO of Google and Alphabet, during the press before the conference. “What all this progress means is that we are in a new phase of changing the AI ​​platform, where decades of research are now becoming a reality for people, business and communities around the world.”

Improved Reflections: The Gemini 2.5 models introduce revolutionary opportunities for “Deep Think”

At the center of Google’s messages is the ongoing evolution of his Twins large language patternswith significant improvements to both Pro and Flash Versions. The updated Twins 2.5 flash It will usually be available in early June, like Pro in a short time after.

The most noteworthy is the introduction of “Deeply think“Improved reasoning mode for the Pro model, which Google claims provides breakthrough efficiency of complex tasks using parallel thinking techniques. The company says that this approach allows the model to look at many options at the same time, like the Alphago revolutionizes the game.

“Deep Think is pushing the model’s work to its borders, giving innovative results,” says Demis Hasabis, CEO of Google Deepmind, during the press. “He gets an impressive result of the USamo 2025, one of the most difficult mathematics indicators. He also leads to Livecodebench, a competition coding standard.”

The company goes on to cautiously with Deeply thinkPlanning to first provide it to the trusted feedback testers before the broader edition. This measured approach reflects Google’s emphasis on the responsible implementation of AI, especially for the borderline capabilities that propel the limits of what AI can achieve.

Search rethinking: AI mode expands with customization and agent characteristics

Google is introducing AI deep into its main search product, unfolding. “AI mode“To all US users, having previously limited it to laboratory testers. This alternative search attempt uses a technique called”Query Fan-Out“In order to break the toll questions and to issue multiple simultaneous searches, to achieve more comprehensive results than traditional search.

“The AI ​​regime is our most powerful search for AI with more modern reflections and multimodality and the ability to go more in-depth through subsequent questions and useful connections to the network,” says Liz Reed, Vice President and head of Google Search.

The company revealed impressive indicators around its existing AI review A feature that now reaches more than 1.5 billion users. “In our largest markets such as the US and India, AI reviews lead to over 10% increase in Google use for types of requests that show AI reviews,” Reed noted during the preview.

The new features coming in AI mode include Deep search For comprehensive research reports, real -time live visuals, and customization options that may include Google Account data to users. This personalization, which requires the explicit inclusion of users, aims to give more appropriate results by understanding individual preferences and contexts.

Google makes a significant boost in shopping, powered by AI, introduces a A virtual feature of experience This allows users to see how clothes will look like them using only one photo of themselves. Technology is a major progress in making online shopping more intuitively and personalized.

“This is a situation where I found perhaps five dresses I like, and I see what it looks like on the website and the models there. However, it doesn’t look like these models and I wonder who will really work for me,” explained the kind of Srinivasan, VP and ADS general manager and trade.

The system is powered by a specialized image generation model designed specifically for fashion applications. According to Srinivasan, he has a “very deep understanding of 3D forms” and fabrics, which allows him to realistically do the way in which clothing items will drap and fit different types of body.

Beyond the visual experience, Google also presents Agency Opportunities that can automatically complete purchases when the items reach a price point determined by the user. This function processes the entire cashier process by Google PayShowing how Google applies its agent AI optimization optimization of daily tasks.

Google has presented significant upgrades to its generative media models by introducing I see 3 for a video generation and Image 4 For images. The most dramatic progress comes in the ability of Veo 3 to generate videos with synchronized audio – including atmospheric sounds, effects and dialogue of character.

“For the first time, we are coming out of the silent era of the video generation,” Hasabis said. “Not only does VEO 3 offer even more fascinating visual quality, but it can also generate sound effects, background noises and even dialogue.”

These sophisticated power models FlowThe new Google AI Movie Creation Tool Designed for creative professionals. Flow integrates Google’s best AI models to help storytellers create cinematographic videos and scenes with a more intuitive interface.

“The stream is inspired by what it feels when the weather is slowing down and the creation is effortless, it is athomatic and full of opportunity,” the company said. The tool has already been tested with several movie makers who have made short films using technology in combination with traditional methods.

Image 4In the meantime, it provides improvements in image quality, with particular attention to typography and text depiction – which makes it especially valuable for creating marketing materials, presentations and other content, which combines visualizations and text.

Immersing Communication: Google Beam develops by Project Starline Research

The company has announced that Project StarlineHis experimental 3D video communication technology was first shown a few years ago, developed in a commercial product called Google BeamS This technology creates the feeling of being in the same room with someone, even when it communicates remotely.

“Google Beam will be a new AI video communications platform,” Picha explained. “Beam uses a new latest AI video model that turns video streams into a realistic 3D experience.”

The system uses an array of cameras to capture different angles of participants, then use AI to combine these streams and to make them on a 3D display in light field with precision head tracking. The result is a deeply immersed conversation experience that goes beyond traditional video calls.

Google partners with HP to bring the first Google Beam Devices for the market of selected customers later this year. The technology also introduces speech translation opportunities that retain the quality and expression of the voice, which allows natural conversations in language barriers – a feature that will also come on Google Meet.

Premium Access: New Ultra Subscription Level is aimed at users and Power professionals

In order to ensure provision of insurance Google have ultrapriced at $ 249.99 per month. This level provides access to the most capable Google models, the highest restrictions on the use and early access to experimental features.

“If you are a director, a developer, a creative professional, or just demand the best of Google AI with the highest level of access, the Google Ai Ultra plan has been built for you -consider it as your VIP pass to Google AI,” said in his press materials.

Ultra plan includes access to Veo 3 with audio generation., Deep -thinking mode When available, Movies., Project Mariner’s agent capabilities and 30TB for storage. It also comes complete with YouTube PremiumS

“The way to think about the Google Ai Ultra plan is that it is almost like your VIP access to all AI on Google. So it will be special features, the highest speed limit. We also put early access to products and features there,” explained Josh Woodward, Vice President of Google Labs and Gemini.

Google’s standard AI Pro subscription of $ 19.99 a month will continue, with some ultra -level features making their way to this more affordable option.

Where research is relevant to reality: Google Ai Vision is shaped

Google’s input/output messages reflect a company at a folding point, successfully turning their huge research investments into products that could change the way people interact with technology. The focus on agents – AI, which can take action on behalf of consumers – signals a significant evolution beyond the current generation of auxiliary AI.

“One of the things I found magical is especially in search of … People are just intuitively adapting to the power of opportunities,” said Pichai. “I think the big thing that people are excited about is when you do (interaction) more natural and intuitive.”

For business and developers who weigh AI strategies, the expanding Google ecosystem offers powerful tools, but requires careful viewing of the roads, costs and consequences of data privacy. The dual approach of the AI ​​installation company in the main products, while developing first -class proposals, implies a long -term strategy to protect existing markets and create new revenue streams.

As these technologies pass from laboratories to daily use, they emphasize the monitoring of Pichai about the present moment of AI: we are witnessing the transformation of theoretical possibilities into practical instruments that meet how people naturally work, create and communicate. The competition for building a truly useful AI is not just about the technical ability -but for bringing intelligence in the moments when we need most, in ways that feel less like the use of technology and more as to understand it.