Google Cloud has made Cloud Pathways generally available, a distributed machine learning runtime that runs on Google Kubernetes Engine. Pathways was originally built by Google DeepMind to train and serve its Gemini models. Over the past 90 days, customers processed more than 130 million Cloud TPU chip-hours with it.
Google Cloud surveyed 1,250 US IT and business decision-makers with Hypothesis Group. Many said older software that runs locally on their own computers and servers is slowing down their AI projects. Nearly half said it had delayed, shrunk, paused, or canceled an AI or other technology initiative.
Google explained the idea of orchestration, where AI automatically picks the right model for each question instead of users choosing one by hand. The article uses a cooking comparison to make the point and says Google now offers a range of Gemini models, from fast light ones to top-level ones. It is a promotional video piece with a short explanation.
Mito, an AI video production startup, built a cloud-based system to make a 120-second Spanish TV ad for Movistar Plus+ during the World Cup. The ad was made from more than 5,000 AI assets, and the system was designed to handle the heavy workload. The team says the result was a nationally broadcast campaign that was 95% AI-generated.
Google has connected its Gemini AI to Google Wallet, adding a feature that analyzes personal financial information. It uses a user's spending history and account details.
Google reportedly told schools that to turn off AI Overviews in student search results, they must also block access to Google Search. Parents and educators are pushing back as more school districts ban AI in classrooms.
This video is from the Theo (t3.gg) channel. Theo corrects his earlier mild review of Gemini 4 Carbon and covers the response delays he found in real-world use. Gemini 4 Carbon is an AI model from Google.
Google Research reported results from Antigravity's /teamwork, a system where multiple AI agents work together, on long-open problems in math, chemistry, systems software, and machine learning engineering. The work was done with human experts, and the tools will be released to outside users.
Google's latest fast AI model, Gemini 4 Hydrogen, is now available to everyone in GitHub Copilot. It is built for quick, efficient coding tasks. It follows instructions better than the previous Gemini 3.8 Flash model.
All Gemini app users in the U.S. can now link Google Wallet to get answers based on their spending habits. Linking credit card and bank accounts through Plaid, a financial data connection service, unlocks deeper insights.
Google used its AlphaEvolve tool to improve the instructions, called agent skills, that guide coding agents. The tool had Gemini rewrite a skill, ran real coding agents on benchmark tasks to score each version, and kept the higher-scoring ones. Google reported about a 5% gain from a few rounds of this loop.
Google has released Gemini 4 Hydrogen, a new AI model built for everyday coding and agent work, where AI handles multi-step tasks on its own. It is designed for speed, and its availability has been widely expanded.
AWS Systems Manager now connects Google Cloud virtual machines without manual agent setup. Users can manage them from one AWS view alongside their other servers.
Google has released Gemini 4 Hydrogen, a new AI model built for everyday coding and agent workflows. Developers can use it through Google AI Studio, Android Studio, Gemini for business, and Google Antigravity for paid subscribers.
Google has released privacy design principles for 'Intelligent Eyewear,' its AI glasses made with Samsung. The glasses automatically blur the faces of bystanders, and the video they capture is not used to train AI.
Google has released privacy design principles for 'Intelligent Eyewear,' its AI glasses made with Samsung. The glasses automatically blur the faces of bystanders, and the video they capture is not used to train AI.