characteristics, price and technical file

by

Asus returns to mobile burden. After launching last year the Zenfone 11 Ultrathe Taiwanese firm has just announced its successor, the Zenfone 12 UltraTerminal bets on the latest Qualcomm, for artificial intelligence and photography to compete in a premium price segment where contestants for Samsung, Xiaomi, Oppo, Alive and company squeeze strongly. We are going to know it better next, but not before it stresses that the device can be achieved in three colors (green sage, ebony black and white Sakura) and that will be available in a single 16 GB version of RAM and 512 GB of storage internal for 1,099 euros. Presentations made, let’s see how this new Zenfone 12 Ultra de Asus. Technical Card of Asus Zenfone 12 Ultra ASUS ZENFONE 12 ULTRA dimensions and weight 163.8 x 77 x 8.9 mm 220 grams screen Amoled Ltpo of 6.78 inches FullHD+ resolution (2,400 x 1,080 pixels) Refresco rate: 1-120/144 Hz Maximum brightness: 1,600 nits HBM 2,500 nits 107.37% DCI-P3 Gorilla Glass Victus 2 processor Qualcomm Snapdragon 8 Elite Adreno 830 GPU RAM 16 GB LPDDR5X Internal storage 512 GB UFS 4.0 rear camera Angular: 50 MP, F/1.9, 1/1.56 “, six axes Great angle: 12 MP, 120º, Free-From lens Telefoto: 32 MP, F/2.4, OIS, 3X optical zoom Video: 8k@30fps Camera in front 32 MP RGBW sensor battery 5,500 mAh Fast charging 65W Wireless load Qi 1.3 Operating system Android 15 connectivity Wifi 7 5G SA and NSA Bluetooth 5.3 NFC Dual GPS Dual nanosim+esim others FUNCTIONS OF IA Dual stereo speakers Survey reader on screen facial recognition Jack of headphones Virtuo dirac Audio and Hi-Rres Wireless Ozoaudio IP68 resistance price 1,099 euros A body so big that even includes Jack Asus Zenfone 12 Ultra | Image: Xataka The Zenfone 12 ultra retains the general lines of the design of its predecessor, but is committed to a more sober finish. Nothing of stripes or decorations, but smooth colors and a smaller camera module that repeats location. We are facing a big deviceof almost nine millimeters thick, and not suitable for weak hands: its weight is 220 grams. More than him Samsung Galaxy S25 Ultrato give an example. The positive part is that this has allowed Asus not to leave anything behind, to the point that the terminal incorporates headphones jack. This port, very popular in past times, will allow us to use wired headphones (wired!). Joking apart, Sound seems to be a strong point Not only because of the double stereo speaker, but for compatibility with wireless hi-res, as well as with Ozo Audio (nokia courtesy audio). With regard to the screen, Asus has opted for Samsung Flex Amoled E6 panel with FullHD+ resolution and an impressive coverage of more than 100% of the DCI-P3 color space. Being Ltpo, admits a Variable soda rate that goes from one to 120 Hzalthough in video games it is capable of reaching 144 Hz. The panel, by the way, is protected by Gorilla Glass Victus 2. On paper, the screen points ways. Bumper from the hand of Qualcomm Asus Zenfone 12 Ultra | Image: Xataka The AI ​​has not taken to reach mobile phones and in the case of Asus it has not been less. The previous model had several functions and this also incorporates them. In addition to the editing and retouching functions, ASUS has implemented transcription tools for voice notes, summary of articles and documents whose execution (based on Call 3 8b of Meta) the Qualcomm Hexagon (the NPU of the processor) is responsible. It also has a call translator and cancellation of calls in calls via AI, I feel the latter the only function that is not in beta. The Asus Zenfone 12 Ultra mounts the Snapdragon 8 Elite, maximum exponent of mobile processors With regard to power, Asus has always stood out for not falling short and In this case it has not been different. The firm has implemented the Snapdragon 8 Elitethe latest Qualcomm, along with 16 generous Gigas of RAM and 512 GB of internal storage. All this gives life a 5,500 mAh battery with a fast -charged 15W and 15W wireless load. According to Asus, it gives for 21.1 hours of content streaming. Nor does it fall short of connectivity. The Zenfone 12 Ultra riding a chip compatible with Wifi 6, 5g and, something very niche but that is always fine to have, Dualsim+Esim. The Bluetooth version is 5.3 and, as far as technologies are concerned, the terminal has everything: Aptx Adaptive and Aptx Lossless Audio of Qualcomm, Dirac, Dirac Virtuo and the aforementioned Hi-Res Audio Wireless Certification. The camera points ways Asus Zenfone 12 Ultra | Image: Xataka Without having the most powerful camera in the market or high -end, Asus Zenfone 12 Ultra promises versatility and good performance, at least on paper. The main camera consists of a sensor Sony Lytia 700 of 50 megapixels with six -aka gimbal stabilizer AKA OIS. The camera, remember, use the technique of Pixel binningso the final resolution will be lower. Next to it is a 32 megapixel telephoto with three increases optical zoom and artificial intelligence. As already do other terminals, making more increases and going to absurd zooms, such as X30, The camera uses AI to improve sharpness and offer better results. The reality is that this varies greatly from one terminal to another and the results are not always perfect. Finally, we have a wide angle of 13 megapixels with 120º of Field of Vision and Free-From lens, that is, a lens that reduces distortion. The selfies, meanwhile, remain in the hands of a 32 megapixel camera hidden in the hole in the central area of ​​the screen. As for video functions, the most remarkable is the recording up to 8kthe stabilization and monitoring of subjects, the equivalent of cinema mode and the reduction of noise through AI during recording. Versions and price of Asus Zenfone 12 Ultra Asus Zenfone 12 Ultra | Image: Xataka He ASUS ZENFONE 12 ULTRA It will be … Read more

2024 has been a year full of uncertainty for chip designers. So much that the market has changed leader

by

2023 was a bad year for companies that are dedicated to the design of integrated circuits. The report published by the consultant Gartner in early 2024 collected that the joint revenues of companies that are dedicated to direct or indirect sales of chips They fell for 2023 11% compared to 2022. The panorama did not paint well by 2024, but there is no doubt about one thing: the year we just left behind has been better for the chips industry as a whole than 2023. According to Gartner During 2024 the income of the semiconductor designers grew 18%. Not bad if we are in mind where they come from. However, this is not at all the only interesting fact that the detailed report that this consultant has prepared. And it foresees that 2025 will also be a very good year thanks to the push of the artificial intelligence (AI). In fact, global revenues will grow, again according to Gartner, of the 626 billion dollars from 2024 to 705,000 million in 2025. Samsung has traced and has been placed again in front of Intel Samsung re -leads the world classification that includes all companies that are dedicated to direct or indirect sale of integrated circuits. Presumably Gartner has not included TSMC in his report, and yes to Intel and Samsung, because the Taiwanese company only manufactures semiconductors for third parties. It does not design or market them directly or indirectly, something that both Intel and Samsung do. In any case, this last company has entered for 2024 66,524 million dollars compared to 40,942 million of 2023. These figures entail a growth of 62.5%. 2024 has been a hard year for Intel greatly due to its difficulties in competing in the hardware market for AI Intel, meanwhile, entered during the year we had just left behind 49,189 million dollars, while in 2023 their income amounted to 49,117 million. It is evident that 2024 has been a hard year For this American company greatly due to its difficulties in compete in the hardware market for AI. The figures that we have just reviewed describe a growth of only 0.1% for 2024 compared to 2023. In fact, this bad economic result led to The company’s departure from the company. If we stick to the gross income Nvidia already steps on Intel’s heels, which has placed it in the third position of the classification. The company led by Jensen Huang entered for 2024 45,988 million dollars, which represents an increase of no less than 83.6% compared to 2023. Behind it has positioned SK Hynix, Qualcomm, Micron Technology, Broadcom, AMD, Apple and Infineon Technologies. Anyway, the most obvious conclusion we can get is that for 2024 the semiconductor market has been promoted above all by the GPUs for AI, and surely this trend is not going to be altered in 2025. Image | TSMC More information | Gartner In Xataka | The virtuous circle: China has become the greatest added value of the planet thanks to feedback

China has built such a huge laser that is seen from space

by

Last year, China revealed a budding technology to which They had called Crazy Li. It was an unpublished combat laser capable of cutting metal or causing blindness on the battlefield. What has now discovered an American satellite is completely different from everything known. By size, we are facing an unusual development, so much that it is seen from space. A technological colossus. Apparently, everything indicates that China is developing in the city of Mianyang, Sichuan province, A huge laser fusion research centera technology with the potential to provide clean and unlimited energy. A project that also could also have military applications in the design and modernization of nuclear weapons, According to analysts from two independent intelligence organizations. And how have we learned? Satellite images have revealed A structure with four gigantic arms that will house laser bays directed towards an experimentation center where fusion tests will be performed using hydrogen isotopes. According to analyst Decker Evelethof the CNA Corporation, this installation is at least 50% larger than the National Ignition Facility (NIF) in the United States, which would make it the largest nuclear fusion center in the world. The company would have been monitoring its evolution from space since in 2020 it was just “a land patch.” The project, called Laser Fusion Major Device Laboratoryit had not been previously reported, which has generated concern in the international community about its true purpose. How the laser fusion works: star energy. Nuclear fusion consists of joining hydrogen atoms to generate energyreplicating the process that occurs in the sun and other stars. Unlike the nuclear fission currently used in nuclear plants, Fusion does not generate long -term radioactive waste or pose catastrophic accident risks. In this type of facilities, the powerful lasers shoot on a central chamber that contains hydrogen isotopes, compressing them until they merge and release a huge amount of energy in a process known as ignition. In the case at hand, the design of the Chinese installation presents those four huge arms that They will channel the laser energy towards a central tower where the fusion chamber is located. China vs.euu. Until now, the National Ignition Facity from the United States has led the laser fusion research for decades, achieving in 2022 a historical milestone: The first fusion reaction with net energy gain. The only “but”: the advance did not tell the energy expenditure necessary to feed the lasers, which still leaves the merger far from being viable as a commercial source of electricity. For its part, China, With its aggressive approach in technological developmentcould be shortening the distances. Analysts such as Melanie Windridge, director of Fusion Energy Insights, highlight that the Asian country is advancing with determination and speedwhich could give it a strategic advantage. For Andrew Holland, director of the Fusion Industry Association, if the United States and its allies do not accelerate the investment in nuclear fusion, China could win the race in a short time. Its ability to carry projects from the concept to construction is remarkably faster than that of the West. China East Fusion Reactor Energy or weapons? It is the big doubt after knowing the images. Beyond its application in energy production, these facilities can also perform A crucial role in the development of nuclear weapons. Both China and the United States are signatories of Complete prohibition treaty of nuclear testswhich prevents testing with atomic explosives. The problem? The capacity of these facilities It allows to simulate nuclear explosions without physically detonating themoffering key data to improve the design and performance of nuclear eyes. In this regard, William Alberque, analyst at the Henry L. Stimson Center, He pointed to the CNN that any country with an installation similar to the NIF inevitably will use its ability to improve its nuclear arsenals. Eveleth coincides, suggesting that the Chinese project It could allow the development of more sophisticated weapons, even smaller and more efficient versions. While some experts believe that this advance reduces the need for real nuclear tests, it could also mean a modernization of the Chinese arsenal without the country needing to perform test explosions, thus avoiding international sanctions. And a hybrid reactor? It is the other possibility that is considered, that the installation in Mianyang is not only a conventional laser fusion installation, but A hybrid fusion-fission reactor, which would further raise its strategic potential. The reason? This type of technology combines nuclear fusion with fission to generate energy more efficiently. According to Holland, if China develops this type of reactor, It would be more advanced than any similar installation in the Westwhich would consolidate its leadership in the field of advanced nuclear energy. Geopolitical implications. China’s advance in the construction of this installation points to part of an ambitious energy and military development program, marking a new phase in technological competition with United Statess. As we said, nuclear fusion has been considered The “energy of the future” for its ability to provide clean and unlimited electricitybut also represents a powerful tool in terms of national and geopolitical security. While the United States continues to lead scientific achievements in the field, China accelerates or seems to be demonstrating A much faster construction and scalability capacity. A development that could influence the balance of global power, not only in the energy field, but also in the strategic and military. A fact to consider: China has performed 45 nuclear tests in its historya significantly lower number than The 1,054 of the United States. In other words, its detonation database is more limited, so the use of advanced fusion simulations could be crucial to improve their designs without the need for real evidence. Image | Planet Labs PBC, Chinese Academy of Sciences In Xataka | His name is Crazy Li and is able to cut metal or cause blindness: China has developed an unpublished combat laser In Xataka | This is ‘Orca’, the new Chinese military ship without crew and up to weapons

We prepare for the cold

by

Beyond The controversy about whether the “eastern beast” It is really a beast or caniche, the models begin to clear many of the doubts we had about next weekend. And they are good news for Cold lovers U bad for everything else. What is the “Beast of the East”? In 2018, A huge temporary ice and snow frozen Much of continental Europe and the British islands. It was then when The name of “Eastern Beast” slipped among the terms used by the media to talk about phenomena similar to that (independently of its intensity). So are we going to a historical stage of cold and snow? No, the truth is that no. It is an “eastern beast” because the mechanism is, more or less, the same: an anticyclonic bridge that connects several areas of high pressures and blocks the casualty movement from the Atlantic to the continent. Basically, we talk about a wall that goes from Scandinavia to Africa following the coastline. That will cause a curious phenomenon: that a mass of polar air moves from the northeast European to the southwest; that is, towards Spain. And what will happen? This type of situation They are difficult to model And, therefore, there are doubts about what their real reach will finally. In fact, now it seems that the dough will not penetrate too much into the peninsular territory. However, that does not mean that you will not put the national meteorology upside down. What we can expect. At the level of rainfall, there are certain doubts: the mass of air comes from central Europe and is very dry. However, on your trip you can generate cold danas or storms that “Incebilize” the situation. The problem is that we will not know until the last moment: it depends on the final configuration, we will have more or less rain. What we have any doubt is that we are about to live a full -fledged winter episode. Nothing historical, not a cold wave: but yesterday nights. These days Let’s suffer some very cold nights in which the majority of the innerior capitals will have negative temperatures. Frost during the weekend They will be extended through the north and peninsular center and can be strong (or very strong) in high areas. What we should not expect. Despite the name, we must not expect a cold wave: neither for intensity, nor by extension, nor by duration. The cold that comes will be anomalous (because winters are getting warmer), but not exceptional. Image | ECMWFF In Xataka | There is something that joins the gigantic calima of 2022 with the nuclear trials of the twentieth century and is called France

Amazon lost the AI ​​train, but wants to recover it. The new Alexa with ia will arrive this month to try

by

Something happens to Amazon. He has been delayed in the AI ​​segment for more than two years. At least that is what seems to the user: it does not have a chatgpt rival, and although it has developed some own models, they do not compete at the moment with those of OpenAi, Google or goal. However, it seems that it will finally try to catch up. There will be news on February 26. We will have to mark that date on the calendar, because it will be when Amazon will finally present its strategy in this area. The company led by Andy Jassy He has sent invitations To various means to notify them that he will celebrate an event with a single protagonist: Alexa. Hello, new Alexa. The event will be presented by Panos Panay, who directed the Microsoft hardware division and will now present the renewed Amazon voice assistant. According to Reuters On February 14, the company’s managers will go to a special meeting to finally decide whether that new version of Alexa is prepared to come to light. Project development It has been chaos according to sources close to it. Waiting for a “remarkable” assistant with a lot of AI. We have been talking about that new version for months that some point to the call “remarkable Alexa” (“Alexa”). The great jump will be provided by the generative AI that will be available in the product and that will theoretically enhance its conversational capacity. Anthropic as a partner? It will be interesting to know which model of AI will be behind the new Alexa. Amazon has invested Friolera of 8,000 million dollars In Anthropic, Claude developer, so it seems feasible to be based on that model. But be careful, because Google too Invested 1,000 million dollars In that company of AI and they also have their own model, Gemini, which could also be an option in the renewed Alexa. There are other possibilities, of course: it is known that Amazon was working In its colossal LLM, Olympusthat it is rumored that it would have two billion parameters, approximately twice as much as GPT-4 of OpenAi. An opportunity for echo. Amazon has a spectacular opportunity here to give a new life to its smart speakers, the Amazon Echo, who were the great excuse to try to take advantage of Alexa. Achieve it, of course, will depend a lot on the quality of the service and if it really represents a substantial improvement of its options. Alexa promised to get us to speak much more with our machines (specifically, with the Echo speakers), but the truth is that few users took advantage of them for something other than establish alarms or ask about time. And for Amazon. Achieving, for example, using Alexa in a “agricultural” way, would make them not only respond to things, but did them for us. They are already able to reproduce a song or a series on the Fire TV Stick, for example, but here the possibilities grow. For example, when combining those AI models with the immense catalog of Amazon products. Ask him to make the purchase or find a certain product and save us thus time (and perhaps money) are some obvious possibilities that could take advantage of this new version of Alexa. Subscribe to Alexa. Rumors suggest that Alexa’s renewed version will be available only through a subscription. That imposes clear doubts About its success, especially when so many other models are available for free and are already very powerful for certain scenarios. If we effectively have a “premium” payment, Amazon will have to have powerful arguments to convince us that it is worth paying for that service. In Xataka | “Telephone, come my life”: Anthropic’s agent wants to change our real lives

The boom and fall of Blackberry, the status symbol that the iPhone turned into history

by

Before the absolute domain of the format imposed by smartphones with any screen without space for the keys, in half the world who left the cod was nokia. In the other half was Blackberry. Its mobile phones, with a reliable QWERTY keyboard and technology to the last, were synonym for quality, even status. Not only public figures had a Blackberry, but their use in business environments was practically mandatory at certain levels, and the company got that His models were desired by the general public. In that area, they had a very special model: the Blackberry Pearl 8100. It was something similar to what is currently happening with the iPhone, but beyond ‘Fardar’, having a Blackberry allowed access to a free instant messaging service among its users. At a time when SMS kept costing money (unless they were included in the rate) and were not as immediate as a chat, Blackberry Messenger It was consolidated as a kind of Microsoft Messenger to talk to friends and family. In 2009, everyone wanted to have a Blackberry mobile and dominated the market with almost 21% share. However, nothing lasts forever and, as soon as the iPhone and the Android mobiles They began to popular, the company’s sales fell into dive. The reason for his success was that physical keyboard that had stopped making sense and to which the brand He wanted to cling without successeven testing hybrid alternatives and Systems like Blackberry 10 To compete against Android. Spoiler: They did not set. And that is only a tip of the Blackberry iceberg, a story that my partner Ana Boria tells perfectly in the video we leave to lead this article, with the perfect storm that led to the loud fall of the mythical Blackberry. And it is curious because, many years later and already in 2025, there are users who continue to remember the fantastic times of that innovative Blackberry and, especially its keyboard and touchpad with those who differentiated themselves from the competition. From outside, as someone who never had a Blackberry, I can only say that it is even sorry for the story of a company that, obviously, was buried by a progress that did not see coming and who did not know how to adapt. Although … well, it’s not like so many other companies that disappeared without a trace. Although their attempts to return to the mobile segment did not work, at least they managed to reinvent themselves in other fields. AND It seems that nothing is going to him. In Xataka | Commodore boom and fall, the giant of the 1980s who revolutionized computer science and fell into oblivion

We already knew that Spanish wine was on its way to collapse. What we didn’t know was that drought was going to accelerate it so much

by

At this point of 2025, say that Spanish wine is on its way to disaster It cannot surprise Nobody. However, it is inevitable that, reading phrases like that, let’s think it is exaggerating. Soon we examine the data, we see that the coup can become huge. Two news that is better understood together. The first is from July 25, 2024: The earliest harvest Within Jerez’s framework since there are historical records. That is to say, For more than 130 years. As the winemakers themselves said, They saw that “In July the grape was already at its optimal point (the 10.5º Baumé demanded) and that if we expected more I was going to lose weight and deteriorate.” The second news is a couple of months later: the production of wine from the United Kingdom has doubled in a very short time and, in fact, the surface planted with vines has increased 75% in the last five years. This is very rare in a place where (despite have vineyards from Roman times and produce commercially since the 60s) the vines have never been good for cold and bad weather. Both news are the beam and the underside of a huge problem: the huge impact that climate change in the main wine regions of Europe has. And, especially, in Spain. A global problem that affects us especially. Traditionally, there are two planetary areas indicated for the cultivation of the vine: the one between the 30th and 50th parallels of the northern hemisphere and that located between the 30th and 40th in the south. The problem is that, like The National Institute for Agronomic Research predicted In France, around 2100 those areas will be completely blurred with the double “very warm” days of the historical average. According to a study published in Nature Reviews Earth & Environment In March 2024up to 70% of the current wine -producing regions could face a substantial risk to lose their suitability for viticulture. In that drawer we are. In Spain, as defends happiness in Herralde, researcher at the Institute for Agrifood Research and Technology of Catalonia, towards the end of the century “the water deficit could reach 200 liters per square meter.” That is, in many wine areas “half of the rainwater that is now available in a year may be missing.” Things are changing. “I have gone from harvesting to do it in a short sleeve and always looking at the sky. My father does not remember in all his youth or a hail storm and now they come to us in September shattering the harvest and even in spring, sweeping that of that year And the next one, because it takes all the yolks “, explained in Rioja2 Berta Valgañónfarmer and producer of the denomination of qualified origin Rioja. And when we say that “time is crazy” we are not entirely aware of what it implies. As Olivia García pointed out “In winter it does not snow, (…) in February it is hot and the plants begin to sprout before but the risk of frosts extends to May (…). In spring it hardly rains and summer is totally dry.” The result is that, when “the harvest arrives so hot that the level of sugar and acidity becomes totally unbalanced.” It is not uncommon. “In a reference period from 1972 to 2005 we have found that, for example, in the Penedès region the increase in average annual temperatures has already reached two and a half degrees,” explained of Herralde in the country. Estimates are terrifying. At the end of 2022, the Reading University published a report where it was concluded that “a fifth of the United Kingdom could have adequate climatic conditions to cultivate Chardonnay grapes in 2050”. But instead, “according to A study conducted over 15 years In vineyards from different areas of the world, 90% of current cultivation areas will not be suitable within a few years. “ To this we must add the problem of water. Not only is water missing at very important specific moments, but As Jordi Pastor defendedmost winemakers already grow with an amount of water lower than optimal. As with the olive tree, the agricultural strategy is to migrate production towards irrigation and, in fact, while 20 years ago the percentages of irrigated vine Today I already touched 50%. And yet, the situation is very complicated … With the available data of the denominations of Catalan origin, we can say that sprouting and flowering are being advanced around 11 days compared to half a century ago. But in addition, “the main cycles of the vineyard (sprouting, flowering, curd, envre and harvest) are faster, those phases are shorter.” Is What we saw too Within the framework of Jerez and In the rest of wine areas from Spain, Greece, Italy or southern California. France, much less affected, too He has seen him The ears to the wolf. … that goes beyond the future. This April, Freixenet presented an ERTE for 615 workers for drought. As they explainedit was an “exercise of responsibility” to “guarantee the operability of the business” in the face of grapes due to the lack of rain. Regardless of the details of that specific case, the truth is that the labor, financial and industrial ramifications The problem is here. And he will not go anywhere. How do we do the wine? “Spain will be a little suitable place to make wine, which means that wine production will not become impossible, but it will be increasingly difficult according to the degree of global warming,” defended Sébastien Zitoresearcher at the Institute of Vineyard Sciences and Burgundy Wine. He is right. Therefore, the world of wine Work already in a hurry for looking for solutions. And the truth is that the struggle to maintain profitability is not the only problem. After all, this environmental pressure also attacks the personality itself of the wines. Can Spanish wines survive while being themselves on the way? Image | Trent Erwin | Climate Resanalyzer In … Read more

All their rivals offer free models that “reason” and Gemini 2.0 is the last example

by

All the companies and startups of AI in the United States were so quiet going to their own. And suddenly Deepseek R1 arrived and became a true existential threat to Silicon Valley. The Chinese startup offered a model of reasoning as good as that of its competitors, but also offered it for free (and Open Source!). What has Silicon Valley did? Apply the story, of course. GEMINI 2.0 Razon Free for All. Enough that you visit the Official Gemini website and display the “Gemini” menu from the upper left to check it. You can already use 2.0 Flash Thinking Experimental (its reasoning model) both in normal mode and in “collaborative” mode with services such as YouTube or Maps. And it is totally free. Microsoft Copilot and Think Deper. Microsoft Copilot’s “Think Deper” mode is also available for free In this service of the company. As we explainThink Deper is actually OpenAi O1, but before Microsoft had to pay the subscription of Copilot Pro ($ 20 per month) to enjoy access to that option. The appearance of Deepseek R1 caused it to also offer it in a grauita way (although with a more limited number of consultations). OPENAI O1. The company led by Sam Altman didn’t want to be left behind and less than a week ago presented O3-minia reasoning model that in addition to being especially powerful is available in the Grauita version of chatgpt. We can activate the “Reason” button so that when we ask something, the O3 reasoning capabilities are put into operation. Deepseek R1 and perplexity. Perplexity’s search engine is gradually offering new options. In fact, a few days ago those responsible announced that On the perplexity website We could activate the Reasoning-R1 model based on Deepseek R1, but housed in the US (to avoid suspicions with possible data theft). They even give the option of opting for the Reasoning-O3-mini model, which is the same offered in Chatgpt. Again for free (although limited), but that stands out for being a comfortable way to try Deepseek R1 in its most powerful version. And the rest? This first batch of reasoning models seems to have taken on foot changed to the rest of the great contenders in the AI ​​segment. Anthropic, who is still a reference with Claude, has not launched a reasoning model at the moment. He has not done so Apple, who goes to his own pace. Meta has not launched anything in this regard despite offering a flame as a clear reference of the Ia Open Source model. And Elon Musk seems to be very busy, because Xai is still working In Grok And for the moment there is no news about a potential variant of reasoning. The only remarkable alternative for the moment is Doubao-1.5-Prothe reasoning model fresh by Bytedance, although it is not available as simple as its competitors. The competition benefits users. The impact of Deepseek R1 on the AI ​​segment has been spectacular as we see. When Openai launched O1 In September 2024 he did it by raising him as a very advanced option but also face: only the subscribers of his services could access it in a limited way. Four months later we are using models that rival O1 but that are totally free and that we can use with more and more options. They are great news for users, which at least for now are benefiting from all that rivalry between these companies. The AI ​​that reasons every time is better and cheaper (or free). A graph Created by Shawn Wang (@swyx) and published in his Newsletter, Latent Space, shows a clear evolution of AI models. In that graph you see how its capacity (measured at LMSYS points, a well -known ranking of AI models) is confronted with its cost per million tokens (ratio 3: 1 entry: exit). Here the right and the right is a model, the better, and Gemini 2.0 Flash Thinking seems to be especially well positioned, but this type of graph is changing very quickly. Again, more good news for us, users. In Xataka | Mistral AI is the French startup that opted for efficiency before Deepseek. His future is uncertain

What are distilled artificial intelligence models and LLM distillation

by

We will try to explain in a simple and understandable way What are distilled models When we talk about artificial intelligence. When we talk to you about Install Depseek on the computer We mentioned that there were distilled versions, and other AIs are also being created that are distilled versions of other specific models. We also usually refer to it as LLM distillationto specify that we refer to Large linguistic models either Large Language Modelwhich are those capable of processing the text, understanding what we write and responding to text. Come on, like Chatgpt , Deepseek, COPILOT, Gemini either Grok. What is LLM distillation The distillation of artificial intelligence models is A technique to reduce the size of the modelsreplicating the results and performance you can get with them. Although we are used to using them through applications and web pages, LLM models They consume a lot of space and resources. We do not usually notice because when you use an AI from a website or app, you connect to the servers of large companies where this model is running. But if you wanted to have a complete model installed on your computer, you would need a very powerful processor and a lot of space. The solution to this problem is to create a distilled model, A model trained to occupy less space. This model can replicate most of the performance, but it will be smaller and fast, you will need less resources to work. The way to do it is similar to a teacher and a student. The complete model is a teacher who shares his experience and knowledge with a student, transmitting complex concepts and knowledge. Meanwhile, the student model learns to imitate what is being taught in a simpler and more effective way. With that, lighter models are achieved. Your results will never be so good like those of the teacher, but the main and performance characteristics will remain. Come on, which comes to be a Lite version, a small but light and versatile version. There are different techniques To create distilled models, such as knowledge distillation with final results for the student model to know the decision -making process or use the teacher to generate additional training data. It is also distilled in intermediate layer so as not to transfer only final results but intermediate layers, or use several teacher models to train the student. In general, private companies that create artificial intelligence models are also responsible for creating distilled versions. The normal thing is that a specific name is added to the distilled version, such as the “flash” of Google Gemini or “Mini” of OpenAi. In other cases, especially In open source modelsThey can use the name of the master model for the distillate but adding as a last name the models that have been used as a student. Come on, you can take a smaller model like Qwen and use it to create a distilled version of Deepseek that is called Deepseek qwen, or Deepseek distill qwen, to indicate that it is distilled. Pros and cons of distilled models A complete artificial intelligence model has billions of parameters, and the quantity of space and computer power To execute them it is huge. In a domestic computer you will need technology and tip power, in addition to a lot of space, already level of a companies such as OpenAi or Google that offer their AI by web or app, you need many resources on their servers. Therefore, creating distilled models helps reduce size and occupy less space. But it also allows them to work faster, and that less computational costs are necessary. That makes Google or OpenAi offering you Free “small” versions Of its main models, leaving the most complete for payment users. Because keeping the complete requires money and investment. And if we are talking about an open source model, have distilled versions allows you and I can install them and use them on our computer without having to spend thousands of euros on a new processor, on graphics cards or internal storage. These techniques can also be used to create artificial intelligence models at a lower cost than would involve complete training. For that, you take already created models and train to a new one from their data and their knowledge, and you do not have to perform the process from scratch. However, distilled models do not have the same amount of data and parameters, they are often lower in resources, and More failures and hallucinations may arise. I will give you an example. If you follow our guide to Install Depseek on the computeryou will see that at a certain point you have several versions. You have versions 8bversions 14bor the full version of 671b. This number refers to its characters, and the lower the less resources you need, but more distilled and small will be the model. Therefore, in this example, if you install an Deepseek 8b and a 14B, you will see that The lower model has more hallucinations And it gives you less precise answers. Therefore, the better you have the greater the model will have to be the model, and less distilled it will have to be. The same goes for commercial models. If you are using a 2.0 flash gemini, the results will be worse than the full Gemini 2.0, and the same with the OPENAI O3 and O3 mini. However, the Flash or Mini version is the one offered to all free users, while the complete is for payment users, in order to assume the cost of maintaining these models in operation. In Xataka Basics | Prompts pages: 16 free websites and communities to find ideas for your prompts and find advice to improve them

Nvidia has a very strong and not quite unexpected ally against Depseek: TSMC

by

Deepseek has convulsed the industry of the artificial intelligence (AI). The irruption of this free and open source Chinese model has questioned the need to use in training and inference processes Chips for very powerful and expensive. Like those designed by Nvidia, which leads with an unappealable roundness the hardware market for Ia. Even so, just ten days ago Deepseek caused Its market value fell abruptly. Since then the hardware used by this Chinese company has generated a lot of distrust. The Deepseek responsible argue that the infrastructure they have used to train their agglutin model 2,048 chips H800 of Nvidia. And that this process with 671,000 million parameters has cost 5.6 million dollars. However, some analysts defend that these figures do not reflect reality. The very juicy report prepared by SEMIANALYSIS He maintains that, in reality, the infrastructure used by Deepseek to train his AI model approximately 50,000 NVIDIA GPU with Hopper MicroAritecture. According to Dylan Patel, AJ Kourabi, Doug O’Laughlin and Reyk Knuhttsen, at least 10,000 of these chips are GPU H100 of Nvidia, and at least another 10,000 are GPU H800. The remaining chips, according to these analysts, are the cuts cut H20. The TSMC cowos packaging is a very strong support for Nvidia As we have just seen, at the current situation it is reasonable to have doubts about the hardware that Deepseek has used in the training of its model (in inference seems to be using the GPUs Ascend 910c of Huawei). And also about the real cost of this process, which could be much higher than the officially announced by this Chinese company. In any case, Depseek has poured uncertainty about the AI ​​market, and this is the reason why so many US technology companies have lost value. Whatever Nvidia has just received a very strong support from TSMC, which is the semiconductor manufacturer that produces its GPU. And is that, according to Digitimes Asiathis Taiwanese company has decided launch an expansion plan five years long for integrated circuit manufacturing capacity using its advanced cowos packaging technology (Chip-on-Wafer-on-Substrate). According to Beth Kindigof the I/O Fund consultant, this technology will monopolize between 50 and 60% of the market in 2025 compared to 15% it supported during 2024. In 2024 TSMC he officially announced that he was building two cowos packaging plants in the town of Chiayi, housed in southern Taiwan The high demand for GPUs for AI with Blackwell MicroAritectura de Nvidia is largely responsible for the implementation of this plan. The company led by Jensen Huang can respond better to the needs of its customers and will see how its competitiveness is increased in a phase in which Depseek and other Chinese companies represent a challenge. In March 2024 TSMC officially announced which was building two cowos packaging plants in the town of Chiayi, housed in southern Taiwan. However, this is not all. He also shuffled the option to put a plant more specialized in this advanced packaging technology in Japan, presumably on the island of Kyushu, in which This company is building at the moment two semiconductor production plants of avant -garde. In any case, there is something else. And it is that Chiayi plants will be trained to work, in addition to the packaging cowos, with advanced Info and Soic technologies (System on Integrated Chips). It is evident that TSMC wants to cover your back well and look to the future to prevent its production capacity from being threatened by a bottleneck. An interesting note: currently the Cowos packaging is being used with the AMD Instinct Mi250 chips and with the A100, H100, H200, B100 and B200 NVIDIA GPUs, as well as in its derivatives. The review used in these last two chips, the B100 and B200, is known as Cowos-L. In 2025 TSMC will be able to process no less than 60,000 wafers per month using its advanced packaging technology. Image | TSMC More information | Digitimes Asia | Yahoo! Finance In Xataka | Samsung is preparing to give TSMC a bars where it hurts most: the manufacture of the chips for ia

Log In

Forgot password?

Forgot password?

Enter your account data and we will send you a link to reset your password.

Your password reset link appears to be invalid or expired.

Log in

Privacy Policy

Add to Collection

No Collections

Here you'll find all collections you've created before.