Gemini 2.5 Deep Think is now available to Google AI Ultra subscribers
According to the company, the latest release incorporates feedback from early trusted testers and research breakthroughs. The new model is a significant improvement over what was showcased at the Google I/O earlier this year. The model is a variation of the one that recently achieved the gold-medal standard at this year's IMO. While the earlier model took hours to reason about complex math problems, Google claims that the latest model is faster and more usable for day-to-day activities. Based on internal evaluations, the new version reaches the Bronze-level performance on the 2025 IMO benchmark.
Google claims that Deep Think could be a powerful tool in creative problem solving. The tech giant gave access to Deep Think on the Gemini app to mathematicians like Michel van Garrel to test mathematical conjectures. Along with Deep Think for Google AI Ultra subscribers, the company is also giving access to the official version of the IMO gold-winning Gemini 2.5 Deep Think model with a small group of mathematicians and academics.
Much like humans who explore different perspectives to a complex problem, weigh in on likely solutions, and revise till the final answer, Deep Think deploys parallel thinking techniques. Google said that this approach allows Gemini to come up with multiple ideas at once and consider them at the same time, sometimes even revising or combining these over time, before offering the best answer.
With Deep Think, Google said it has extended the inference time, or thinking time, endowing the model with the time to explore different possibilities and come up with creative solutions to complex problems. Besides, the company has also developed novel reinforcement learning techniques that push the model to use extended reasoning paths, allowing it to become a better problem-solver over time.
When it comes to use cases, according to Google, Deep Think can help users tackle problems that need creative thinking, strategic planning, and making improvements in a step-by-step manner. When it comes to design development, the model can improve both the aesthetics and functionality of web development tasks. It can be useful in scientific and mathematical discovery, as this model can assist in formulating and exploring mathematical conjectures and can even reason through complex scientific literature. Since Deep Think is believed to excel at tough coding problems, it can be of help in algorithmic development and code.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Indian Express
13 minutes ago
- Indian Express
Silicon Valley Is in Its ‘hard tech' era
In a scene in HBO's 'Silicon Valley' in 2014, a character who had just sold his idea to a fictional tech company that was a thinly veiled analogue to Google encountered some of his new colleagues day drinking on the roof in folding lawn chairs. They were, they said tipsily, essentially being paid to do nothing while earning out — or 'vesting' — their stock grants. 'Rest and vest,' the techies said, in between sips of beer. The tongue-in-cheek send-up wasn't far from Silicon Valley's reality. At the time, young engineers at Facebook, Apple, Netflix and Google made the most of what was known as the 'Web 2.0' era. Much of their work was building the consumer internet — things like streaming music services and photo-sharing sites. It was a time of mobile apps and Mark Zuckerberg, Facebook's founder, wanting to give everyone a Facebook email address. It was also the antithesis of corporate America's stuffy culture. Engineers held morning meetings sitting in rainbow-colored beanbags, took lunch gratis at the corporate sushi bar and unwound in the afternoon with craft brews from the office keg (nitrogen chilled, natch). And if they got sweaty after a heated office table-tennis tournament, no matter — dry cleaning service was free. That Silicon Valley is now mostly ancient history. Today, the tech has become harder, the perks are fewer and the mood has turned more serious. The nation's tech capital has shifted into its artificial intelligence age — some call it the 'hard tech' era — and the signs are everywhere. In office conference rooms, hacker houses, third-wave coffee houses or over Zoom meetings, knowledge of terms like neural network, large language model and graphical processing unit has become mandatory. Stacked up against ChatGPT's ability to instantly transform any image into a Studio Ghibli cartoon, Instagram's photo filters are practically Paleolithic. And the chatter is about not how you built your app with the HTML5 coding language, but how many H100 graphics cards — the highly coveted hardware for running AI programs — you can get your hands on. The tech epicenter has moved from the traditional cradle of Silicon Valley — the towns of San Jose, Mountain View, Menlo Park and Palo Alto — 40 miles north to San Francisco, the home of the AI startups OpenAI and Anthropic. Tech giants like Google are no longer hiring in droves as they once did. And those with jobs at those behemoths are met by the watchful eyes of managers looking to cut dead weight rather than coddle employees. The region, long known for its capital-L Liberal politics, is no longer a political monoculture. A contingent of venture capitalists and entrepreneurs has spurred a rightward shift, leading to the rise of the 'Liberaltarian' — a term coined by two Stanford political economists to describe the tech industry's proclivity toward trumpeting liberalism in some social issues but maintaining anti-government posturing in regulating businesses. Alongside that change, industries that were once politically incorrect among techies — like defense and weapons development — have become a chic category for investment. If Silicon Valley's Web 2.0 era was defined by founders playing God on their computers by creating social networks and other services, the new era is about founders angling to create 'superintelligent' computers that may one day surpass humans and become a kind of 'God' in the machine. 'The low-hanging-fruit era of tech, where earlier consumer-facing software businesses were easier to build and printing money, it just feels over,' said Sheel Mohnot, a general partner at the venture capital firm Better Tomorrow Ventures in San Francisco. Silicon Valley has long provided a peek into the possible future, and the region's changes over the past 15 years give a glimpse into the technologies and corporate cultures that may spread across industries and, eventually, across the rest of the world. After some years of dizzying highs (Twitch! Waymos! Crypto!) and very low lows (Metaverse! Clubhouse! Crypto!), we would do well to stop and see how the nation's tech capital has altered. We might learn something about how we will soon live. I've seen these changes firsthand, having lived in the San Francisco Bay Area for 20 years and worked as a tech journalist since 2010. I began chronicling Silicon Valley in its Web 2.0 heyday, when Facebook and Google — still in their adolescence — were replacing older companies like Sun Microsystems as the cool places to work. In some cases, quite literally. Facebook bought Sun Microsystems' corporate campus in Menlo Park in 2011, making it the headquarters of a social networking empire that absorbed Instagram and WhatsApp. Facebook, later renamed Meta, also commissioned Frank Gehry to design a new postmodern office complex next door. Down the road, the Googleplex in Mountain View was sucking up talent and growing like a weed. Larry Page and Sergey Brin, Google's founders, codified the now famous '20% time' perk, allowing employees to work on whatever they wanted one day a week. 'Google kicked off the lunches and the nap pods and all that stuff, and everyone else just had to compete,' Mohnot said. With interest rates for borrowing cash at rock bottom in the late 2010s and early this decade, Web 2.0 startups raised money easily. One company, Zume, raised nearly $500 million to revolutionize automated pizza making. Then came the pandemic. Some techies who worked in Silicon Valley but lived in San Francisco were fed up with what they saw as the city government's getting too 'woke.' Others saw an opportunity to work remotely. Many decamped to Miami, Austin or Jackson Hole. Downtown San Francisco hollowed out. Facebook, Google and others hired aggressively in 2020 and 2021, as people were trapped indoors and their internet use soared. But after lockdowns lifted, the companies found they had recruited too many people. In 2022, Zuckerberg slashed roughly a third of Meta's workforce. Elon Musk, who bought Twitter, now known as X, shed three-quarters of its employees. Other companies followed, eliminating moderators, marketers, media handlers and all things associated with diversity and inclusion. Heaven help those with a humanities degree. That same year, OpenAI — a relatively obscure AI lab in San Francisco — unveiled ChatGPT, which exploded in popularity. Suddenly, AI was the hottest thing. Meta, Google and others joined the AI rush, shedding 'softer' skilled employees for 'harder' ones. Digital prophets, whose jobs amounted to being full-time TED Talk deliverers, were out. Deep learning and neural network specialists were in. Venture capitalists, undaunted by failed investments in crypto and metaverse startups, threw their money at any entrepreneur with 'AI' or 'machine learning' in the pitch deck. As OpenAI, Anthropic and other AI startups gained prominence in San Francisco, entrepreneurs abandoned building a 'Silicon Beach' in Miami or a 'Silicon Hills' in Austin. They returned to the Bay Area. 'For something so up in the cloud, AI is a very in-person industry,' said Jasmine Sun, a former employee at Substack who is a culture writer in San Francisco covering tech and AI trends. 'People are going to parties, hacker houses, happy hours. It's where they all mingle and exchange information.' The AI influx has remade parts of San Francisco. The space between the city's Mission District and Potrero Hill neighborhood is now called 'The Arena' after startups like Notion and Chroma moved in along with OpenAI. Like Russell Crowe in 'Gladiator,' founders in 'The Arena' are fighting one another for AI supremacy, though on the battlefield of free-market capitalism and not in the Colosseum. Hayes Valley — an edgy neighborhood that has gentrified — is called 'Cerebral Valley,' a hot spot for AI engineers to meet over warm lamb and harissa salads at Souvla. Nearer the water, the Ferry Building on the Embarcadero is home to Shack15, a private coworking space filled with startup workers eager to network in between vibe-coding sessions. As a more right-leaning crowd of tech elite and terminally online Gen-Z founders emerged, they turned against politics in the workplace and globalism. Sharing technology globally had put Silicon Valley's tech leadership at risk of being overtaken by China, some said. New startups reflected the evolution. Instead of tap-to-pay apps of a decade ago like Clinkle and Bump, young companies making unmanned aerial drones stocked with AI-guided Barracuda cruise missiles appeared. Think of the change as less HBO's 'Silicon Valley' and more HBO's 'Mountainhead.' Still, an appetite for exuberance remains. Roon, a computer scientist and well-known member of the tech commentariat on X who is known by his handle, recalled a party around the time of ChatGPT's 2022 release as a watershed AI moment for San Francisco. Stability AI, a startup that makes image-generation software, rented the city's Exploratorium science museum — founded by Robert Oppenheimer's younger brother, Frank — to celebrate $101 million in funding. 'It was fun, it was real, the tech was real,' said Roon, who declined to be named because he works for a prominent tech company but does not represent it. There is a sense of cautious optimism — even renewal. Uber subleased 500,000 square feet of office space to OpenAI in 2023, as the company expanded. Scale AI, which Meta recently invested $14.3 billion in, took over a building in San Francisco's Design District that had been occupied by Zynga, the gaming company. Even the once sleepy Financial District is bustling again. 'We are so back,' Roon said.
Hindustan Times
16 minutes ago
- Hindustan Times
Google brings iPhone-style swipe animations to Android browsers: Here's how to turn them on
Google is bringing a new navigation feature to Chrome on Android that closely resembles iPhone-style swipe animations. The update introduces a page transition effect, which gives users a preview of the previous or next screen when performing back or forward swipe gestures. Here's how to enable iPhone-style swipe animations in Chrome for smoother back and forward navigation.(Pexels) The feature builds on Android's Predictive Back system, which was first introduced with Android 13, allowing users to see the destination screen before completing a back gesture. In Chrome's latest test, this functionality extends to browsing, showing a dimmed preview of the previous page within a tab's history. If the gesture is performed on a new tab without history, Chrome displays a grey background with its logo, indicating a return to an empty tab. Also read: Xiaomi Unveils New AI Voice Model to Boost Auto, Home Tech According to reports from Android Authority, the animations are being tested in Chrome version 138. While the feature was previously hidden behind developer flags, it is now appearing for select users in the public build. How to Enable Chrome's Swipe Animations on an Android Phone Open Chrome on your mobile device and type chrome://flags in the address bar. Use the search function to locate the flags #back-forward-transitions and #right-edge-goes-forward-gesture-nav. Enable both flags. Chrome will then display a 'Relaunch' button at the bottom of the screen. Tap it to restart the browser and activate the animations. Also read: CEO Tim Cook says Apple is ready to open its wallet to catch up in AI As an alternative, you can type chrome://flags#back-forward-transitions and chrome://flags#right-edge-goes-forward-gesture-nav directly into the address bar to enable them without searching. Once restarted, Chrome will show the new swipe animations when navigating between pages. Google has not enabled this feature by default, and since it is still experimental, some visual glitches may occur. Users can disable the flags by following the same steps if they encounter issues. Also read: Netflix quietly used AI for scenes in its new show: Here's why it could change what you see on screen The move aligns Chrome's gesture navigation more closely with the fluid animations found on iOS, giving Android users a more interactive browsing experience. While the feature is still in testing, it signals Google's intent to refine touch-based navigation for its mobile browser.
The Hindu
an hour ago
- The Hindu
Google agrees to curb power use for AI data centres to ease strain on U.S. grid when demand surges
Google has signed agreements with two U.S. electric utilities to reduce its AI data centre power consumption during times of surging demand on the grid, the company said on Monday, as energy-intensive AI use outpaces power supplies. Utilities in the country have been inundated with requests for electricity for Big Tech's AI data centres, with demand eclipsing total available power supplies in some areas. That power crunch has led to concerns about spiking bills for everyday homes and business and blackouts. It has also complicated the technology industry's expansion of AI, which requires massive amounts of electricity, fast. Google's agreements with Indiana Michigan Power and Tennessee Power Authority would involve scaling back power use at the technology giant's data centres when called upon by the electric utilities to free up space on the grid. They are the first formal agreements by Google in demand-response programmes with utilities to temporarily curtail its machine learning workloads, a subset of artificial intelligence. "It allows large electricity loads like data centers to be interconnected more quickly, helps reduce the need to build new transmission and power plants, and helps grid operators more effectively and efficiently manage power grids," Google said in a blog post. Demand-response programmes have typically been used by other energy-intensive industries like heavy manufacturing or cryptocurrency mining. In exchange, the businesses generally receive payments or reduced power bills. The programmes involving AI activity in data centres is generally new, and details of the commercial arrangements between Google and the utilities were not clear. While demand-response agreements apply only to a small portion of demand on the grid, the arrangements might become more common as U.S. electricity supply tightens.
