powerful

AI chips have always wanted to become more and more powerful. TSMC has just pointed out the true limit: efficiency

by

More performance? It is the first thing we usually ask of a new chip, almost without thinking about it. We have done it for years with the processors in our devices and we do it now with the chips that support much of the deployment of AI. More computing power, more speed, more scope to do things that previously seemed out of reach. But this logic begins to encounter a very specific limit: energy. What is making its way now is a less flashy idea, but increasingly difficult to ignore: progress will not only be measured by how much a chip calculates, but also by how much energy it needs to do it. The clearest clue comes from TSMC. We are talking about the largest contract chip manufacturer in the world, a company that does not sell processors under its own brand, but rather produces semiconductors designed by other players in the industry. According to ReutersKevin Zhang, senior vice president of business development, explained at a conference in Amsterdam that his customers are paying more and more attention to performance improvements that do not increase consumption. The pressure comes from very different profiles, from smartphone manufacturers to AI data center operators, all with a concern that we have seen growing in recent times: electricity cost and energy availability. The key is in the manufacturing. TSMC has not simply described a change in priorities. He has also placed it on his technological calendar with A14a future manufacturing technology planned around 2028. The firm expects that this process offers more than a 20% improvement in performance and, at the same time, reduces consumption by up to 30% compared to N2, the process that the company takes as a reference in that comparison. The key is that we are not talking about a specific processor, but rather the method with which subsequent chips can be manufactured. Not everything is about miniaturizing. For decades, reducing the size of transistors has been one of the great ways to gain performance and efficiency in chips. That logic doesn’t go away: transistor density remains within TSMC’s roadmap. What Zhang points out is that in the face of energy pressure from AI, other solutions, such as advanced packaging, chip stacking, and photonics, are also gaining weight. In parallel, as we pointed out a few weeks agoTSMC has decided not to use High-NA EUV, the lithography associated with ASML’s most advanced and ambitious equipment, in its A13 and A12 processes planned for 2029. The battle is also in the data. Huawei enters this conversation with Tau Scaling Lawa proposal that seeks to improve performance by accelerating the movement of data within the chips. The idea shifts part of the focus from the transistor to architecture and integration, two areas that gain weight when manufacturing smaller components is not enough. Along the same lines appears LogicFolding, which Huawei presents as a possible step beyond traditional 3D stacking, but which will depend on new design tools for folded architectures and better dissipation solutions for devices ranging from smartphones to AI data centers. Where are we going? TSMC does not speak for the entire industry, but its position makes the message carry. The firm suggests that, at least in its roadmap and in conversations with its clients, energy efficiency is gaining prominence that was previously more hidden behind performance. And it’s not a concern limited to AI data centers. Huawei, for its part, shows that the problem is also being addressed from architecture and integration, not just from the manufacturing process. The common point is not a closed conclusion, but an increasingly visible tension: chips will have to continue to be more capable, but each leap will be more difficult to justify if it increases consumption, heat or costs. Images | Xataka with Nano Banana In Xataka | Samsung has just achieved a milestone that has not been recorded for eight years. The problem is that it is a mirage

Tenerife was known for the sun and its beaches. It will soon house one of the five most powerful supercomputers in Spain

by

Tenerife will have a new supercomputer. I already had two with the names of Teide and of Anagaand they will now be joined by a new and promising project called the Atlantic Supercomputing Center. With it, it is hoped to turn the Canary Islands into a new nerve center for retaining and attracting talent in the technological field. Up to 10 million euros of investment. This new project It is a collaboration of the Cabildo of Tenerife and the Institute of Technology and Renewable Energies (ITER) with the German technology giant Bechtle. It will have an initial investment of 5.5 million euros, which could rise to 10 million as its four phases are deployed (two for storage, two for computing) oriented by the demand for the center and its resources. The expansion is flexible and Bechtle will supply the latest technology available at the time of project execution to avoid the use of obsolete components. The fifth supercomputer by power in Spain. By integrating with the existing nodes, the Atlantic Supercomputing Center will achieve a combined power that will place it as the fifth most powerful supercomputer in the entire national territory. It is also expected to enter the prestigious TOP500 list which brings together the most powerful supercomputers from around the world. Hybrid architecture. The rise of AI has meant that the project has an architecture that will allow working with both more conventional workloads and those intended for projects in the field of artificial intelligence. Thus, its architecture will be hybrid: CPU: although it has not been specified which processors it will use, it has been indicated that the supercomputer will have 13 nodes with 288 cores each, which will allow for more than 3,000 process cores to execute scientific tasks, for example. GPU: there will also be four specialized nodes with a total of 32 Nvidia H200 NVL cards, which will allow training of large language models and the development of AI projects. Performance: this expansion is expected to provide between 1.3 and 1.4 PFLOPS of global computing power (close to 300 TFLOPS in CPU and almost one PFLOP in GPU), indicated those responsible for the Cabildo de Tenerife and ITER. Hours instead of months. The president of the Cabildo, Rosa Dávila, stood out that local laboratories, the University of Las Palmas de Gran Canaria and the University of La Laguna among others will be able to access these resources to be able to compute in hours what previously could take months. Juan José Martínez, from ITER, recalled how during the pandemic the Teide-HPC supercomputer It was one of the five centers in all of Spain who sequenced and monitored the biological variants of COVID-19. From the audiovisual sector to the aerospace sector. Among the sectors that will benefit from this computing capacity will be those associated with the audiovisual industry. The Teide-HPC infrastructure was for example used to render scenes from the film ‘Tadeo Jones 2: The Secret of King Midas‘. It will also be the core of the project management of canary satellite constellation. Attracting talent. This facility also wants to become an element that reinforces the role of the Canary Islands as a technological hub. Having a supercomputing infrastructure like this wants to help attract technology companies that promote highly qualified young employment and therefore retain and attract new talent in this sector. Efficiency. Although the power of Teide HPC will greatly benefit from these new resources, advances in photolithography will mean that the new supercomputer will occupy only a quarter of the previous physical space. Its environmental impact will also be zero: the infrastructure will be located in ITER’s own facilities, and will be powered entirely with clean energy from its wind farms and photovoltaic plants. Image | POT | ITER In Xataka | The muscle of many supercomputers depended on GPUs: China is trying another way to surpass the best in the US

China is about to launch the most powerful cargo drone in the world. And it will move it with hydrogen

by

The aeronautical industry has been researching and experimenting for quite some time. with hydrogen turboprop engines on airplanes. A Chinese company is about to break that barrier, as it has taken off an airplane with one of these megawatt-scale engines. Aero Engine Corporation of China (AECC) has completed the first test flight of the AEP100, installed on a 7.5-ton cargo drone, in an operation that took off from Zhuzhou airport, in Hunan province. what has happened. The device flew for 16 minutes, reached an altitude of 300 meters and traveled 36 kilometers at a speed of 220 km/h before landing without incident. According to AECC, the engine operated stably throughout the flight profile and responded as expected. Chinese state media present it as the world’s first flight with a hydrogen turboprop of this power. Why is it relevant?. Yes, it is a short, unmanned, low-altitude test. But this means that hydrogen aeronautical propulsion leaves the laboratory and test benches to face real flight conditions. AECC maintains that the country already has a complete technological chain for hydrogen aeronautical engines, from essential components to system integration. direct combustion. The AEP100 does not use fuel cells to power an electric motor. It burns liquid hydrogen directly in a turbine cycle, just as a conventional turboprop would burn kerosene. This is the main difference with other bets. Airbus, for example, has prioritized fuel cells on its roadmap to a hydrogen commercial aircraft in 2035, while China has opted for direct combustion. Combustion is more complicated to tame in engineering, but offers much higher power density, something key to scaling up to larger aircraft. What aircraft is it intended for?. The AEP100 is custom designed for the W5000, a twin-engine cargo drone developed by Chinese startup Air White Whale. According to the manufacturer’s data, we are talking about a device with a maximum takeoff weight of 10.8 tons, 5 tons of payload, more than 65 cubic meters of hold and a range of 2,600 kilometers. Just like share from China Daily, when it completes its first flight, it will become the most powerful transport drone in the world, surpassing the Norinco Luca. Deadlines. Yuan An, general manager of subsidiary AE General Aviation Power Tech, has explained The engine is in the final phase of the type certification process and they hope to obtain approval from the Civil Aviation Administration of China in 2027. The process is progressing faster than usual because the AEP100 shares a core with the AES100 turboshaft, which shortens procedures. Yuan has also assured that the AEP100 and its variants will “end the heavy dependence on foreign engines” in Chinese general aviation. Where will it be used first?. For now, we have to forget about getting on a hydrogen-powered passenger plane. The bet goes through what they call the “low altitude economy”that is, situations in which unmanned cargo drones, inter-island logistics or controlled transport routes to remote areas are used, being scenarios where hydrogen refueling infrastructure, certification and operational economics are more manageable than in passenger aviation. Yuan remember also that the United States has more than 275,000 general aviation aircraft, while in China there are only a few thousand. The problems that remain unresolved. Burning hydrogen in a turbine is no small feat, as you can imagine. It burns at higher temperatures than kerosene and with a much higher flame speed, which requires the design of systems that avoid autoignition, flame flashbacks and combustion oscillations. Added to this is storage, since liquid hydrogen requires cryogenic temperatures close to -253 ° C, heavily insulated tanks and, most likely, redesigning the geometry of the fuselage itself to accommodate it. Sustainability. aviation Today it is around 2% of global CO₂ emissions, a figure that could skyrocket in the coming decades if the sector maintains its dependence on fossil fuels. China aims to reduce its exposure to imported oil in an increasingly complicated geopolitical scenario, so hydrogen can fit into both narratives. And now what. China’s road map mark 2028 as horizon to validate similar technologies in small unmanned aircraft, helicopters and urban air mobility, 2035 for applications in broader regions and 2050 for large commercial turbofan aircraft. The first flight of the W5000 with the AEP100 installed is expected in the coming months and will be the next litmus test. Cover image | CCTV In Xataka | For China, DeepSeek is more than just AI: it is the key to creating an industry that makes them independent of Nvidia

NASA prepares chips 100 times more powerful

by

Human beings explore because they need to understand what lies beyond. We have done it by crossing oceans, climbing mountains and, for decades, sending machines to places where we cannot yet be. But a space mission has more to do than get there. For example, collecting data, interpreting it and sending it to Earth to do science. This is where the great challenge appears, because space requires computers capable of functioning for years in an environment that punishes electronics like few others. High Performance Spaceflight Computing. This is the name of the response that NASA is preparing. According to the agencythe project seeks to develop together with Microchip Technology a new space processor capable of offering up to 100 times more computing capacity than current space computers. We are not talking about a chip designed for a laptop or a mobile phone, but rather a system on a chip, or SoC, called to be integrated, once certified for space flight, in future ships, orbiters, rovers, manned habitats and deep space missions. SoC, a familiar term. This is the type of architecture that is common in our smartphones and tablets: small devices that concentrate essential elements of a computer in a single piece. The difference compared to an isolated processor is precisely there. An SoC is not limited to executing instructions, but can integrate CPUs, computing support units, advanced networks, memory, and input and output interfaces. On Earth we use it to gain efficiency and reduce size. In space, moreover, it has to survive. The challenge. As we say, space punishes electronics in a way that we rarely see down here. According to NASA, a processor intended for real missions must withstand electromagnetic radiation, extreme temperature fluctuations and high-energy particles capable of altering the operation of the systems. We are not just talking about losing performance, but about errors that can force a ship to enter “safe mode”, with non-essential operations turned off until mission teams resolve the incident. A key phase. Now comes the time to check if everything that is promised on paper holds up when taken to the physical field. JPL began testing in February and will maintain the campaign for several months, with radiation tests, thermal cycles, shocks and functional evaluations. The agency ensures that the processor is working as designed and adds a striking fact, although still within the framework of the tests: the first indications show that it operates with a performance 500 times higher than the radiation-hardened chips currently in use. More autonomy away from home. Space exploration has a limit that is not resolved with a larger antenna: distance. Between Earth and Mars, ua signal may take a while between 3 and 22 minutes to travel in one direction, depending on the position of both planets in their orbits. That means we can’t drive a rover like someone drives a remote-controlled car. We have seen it in the Martian landings, the famous “seven minutes of terror”, when a ship enters, descends and lands, executing a choreography by itself that from Earth we can only know when it has already happened. On-board computing. NASA proposes that this type of processor will allow future ships to use artificial intelligence to respond in real time to complex situations, analyze large volumes of data, store it and transmit it more quickly. Let’s remember the case of Perseverance which already combined orbital data of Mars, its panoramic camera and a Snapdragon 801 to compare what he saw with information obtained from space and refine his position on the Martian surface. If we want to continue exploring Mars and look further, we will need more and more systems capable of making decisions without always waiting for an order from Earth. Technology that returns. The history of space exploration is also the history of ideas that are born to solve very specific problems and then find a place on Earth. In this case, NASA points to possible adaptations for sectors such as aviation and automotive, in potential uses such as drones, electrical networks, medical equipment, communication services, artificial intelligence and data transmission. It does not mean that we will see this processor in a consumer product tomorrow, but it does mean that the effort to make it more powerful, efficient, scalable and resistant can go beyond a ship on its way to deep space. Images | POT In Xataka | The biggest problem with living on the Moon is its nights. NASA believes it has found the solution to avoid running out of electricity

The energy jets from black holes are so powerful that they can reshape entire galaxies and now we know how to measure it

by

It is always said that black holes They gobble up everything that comes close to themfrom matter to light. However, this is not entirely true. In some cases, there is a fraction of particles and energy that, instead of falling inside, does the opposite. It is ejected in the form of jets, known as jets. Although there are some hypotheses about this, the reason why this occurs is not completely known. What is known is that these jets are so powerful that they can even influence the evolution of galaxies. The problem is that it is known that they are very powerful, but not how powerful. Until now, no one had been able to directly measure the power of these jets. However, an international team of scientists has achieved measuring these jets around a specific black hole, thereby opening up a very interesting range of possibilities. The data. These scientists have studied the Cygnus X-1 systemcomposed of a black hole and a blue supergiant star orbiting each other. Using a very novel method, they have discovered that the energy of the jets leaving the black hole is equivalent to that of 1,000 suns. They have also observed that they move through space at a speed of 540 million kilometers per hour and that 10% of the energy that is initially formed in the fall towards the black hole is converted into jets. The background. Until now, no one knew how to measure the power of a black hole’s jets. The only thing that was done was to measure the scars they left in space using calorimetric methods. When freed, they can leave in their wake hot spots and holes in the intergalactic medium. However, As explained in an article by Interesting Engineeringthis is something like wanting to measure the power of an engine by observing the treads of the car’s tires. The important thing is to directly analyze the machinery. And that is precisely what has been achieved now. Indirect measures. In systems formed by a black hole and a star, the black hole feed little by little gas surrounding the star. As it approaches it, the gas begins to rotate faster and faster, generating a lot of heat and energy. Part of that energy does not fall into the black hole, but instead jump outward, forming the jets. In turn, the star releases very intense flows of particles, which give rise to what is known as stellar winds. Those stellar winds can interact with the jets and bend them. And there is the key. The jets cannot be measured as such, but the resistance they offer to being bent by stellar winds can be measured. For example, we can know how strong a person is by analyzing his or her ability to beat someone whose strength we do know in an arm wrestling match. Trajectory changes. The overall trajectory of the jets depends on the momentum flux of both the jets themselves and the winds. Since the momentum flow of the wind can be calculated, it is enough to analyze the trajectory to solve the unknown. The data can also be further refined with a series of computer simulations. The result is a fairly rough estimate of the power of the jets. There are limitations. The biggest limitation of this study is that only one black hole has been analyzed. The procedure would have to be repeated with more jets in more black holes to check if there is a trend and, therefore, if the method is valid. Galactic evolution. Since jets from larger black holes can significantly affect galactic evolution, this method could be very useful to better understand how galaxies form. That is why it is important to move on to the second step and check if the method is reproducible, especially with larger black holes. Image| A supermassive black hole ejects a jet of plasma 3,000 light years long, traveling at almost the speed of light. NASA artist concept In Xataka | We thought that the heart of the Milky Way was an immense black hole. Mathematics has changed this idea for us

China is preparing the most powerful and rare exascale supercomputer on the planet. No GPU: only Chinese CPUs

by

An exascale supercomputer is one capable of performing at least 1 exaflop (10¹⁸) of floating point operations per second. These machines are the most powerful currently available if we stick to classic computers and leave aside the prototypes of quantum computers. The classification TOP500 identifies the most capable supercomputers on the planetand, as expected, four exascale machines appear at the top of this list: The Captain, FrontierAurora and Jupiter. The first three reside in the United States and the fourth in Germany. Curiously, no Chinese supercomputer appears in the top ten positions of this classification, although we know that some of its most powerful machines are not officially reported to the TOP500 for geopolitical reasons. Be that as it may, the Government led by Xi Jinping is determined to change this scenario. And the Shenzhen National Supercomputing Center has announced that is going to build a supercomputer called Lingshen that, according to this institution, will have a sustained performance of more than 2 exaflops and will integrate only components designed and manufactured in China. Lingshen supercomputer architecture is very unusual The supercomputer ‘The Captain’ from the Lawrence Livermore National Laboratory (USA) is a real beast. This machine exceeds 1.8 exaflops, making it currently the most powerful on the planet. The APUs are responsible for its brute force. Instinct MI300A from AMD, which work hand in hand with the EPYC 9005 processors. However, the most surprising thing is that it brings together no less than 11,340,000 cores and delivers 1,809 PFlops/s Rmax and 2,821.10 PFlops/s Rpeak. Lingshen will bring together 47,000 processors of Chinese origin that will be distributed in Huawei Kunpeng servers The architecture of ‘El Capitan’ is very similar to that of the other supercomputers in the TOP500 classification, but the machine being prepared by the Shenzhen National Supercomputing Center is going to take different paths. And it is that according to Lu Yutongthe director of this center, the Lingshen supercomputer will use only general purpose processors (CPU), and will not use GPU. Not a single one. It is a very unusual decision, and it is surprising that in theory it will exceed 2 exaflops only with this type of chips. Be that as it may, this is not the only thing we know. Lingshen will bring together 47,000 processors of Chinese origin that will be distributed in servers Huawei Kunpeng equipped with Taishan cores with ARM architecture. Lu Yutong has also confirmed that this machine will have 650PB of storage and a million-port interconnection. Everything that the Shenzhen National Supercomputing Center has announced sounds great, but this project also leaves us with some very reasonable doubts. The most obvious is that Lingshen is just a project at the moment. It has not yet been built, so its theoretical maximum performance comes from an estimate and not from a measurement provided by a real test bench. On the other hand, it is very surprising that the Shenzhen National Supercomputing Center has chosen to integrate only CPU. Huawei, Moore Threads and Cambricon Technologies are three of the chinese companies which have domestically made GPUs that could presumably fit into this machine. In any case, it is worth keeping track of this project to see if Lingshen finally lives up to the expectations it has raised. Image | TOP500.org More information | Shenzhen National Supercomputing Center In Xataka | The Frontier supercomputer is the second most powerful exascale machine on the planet. And it has a mission: nuclear fusion

In two days the animated spin-off of the platform’s only powerful franchise premieres on Netflix: ‘Stranger Things’

by

On December 31, 2025, Netflix aired the final episode of ‘Stranger Things‘. With it, the platform recorded its most viewed New Year’s broadcast in history and put the finishing touch to the platform’s most viewed series, exceeding 1.2 billion accumulated views. And four months later Eleven, Mike, Dustin and the rest of the Hawkins gang are back, but this time in animated format With new voices and a new showrunnerthis ‘Stranger Things: Stories from 85′ this one arrives April 23 to Netflix. And how is it presented when the main story has already closed? Well, going back to when it wasn’t yet: the series takes place in the winter of 1985, between the second and third original seasons. That interval existed in the canon, although it did not come without a technical problem: at the end of T2, Eleven closes the door to the Other Side. T3 starts in July of that same year. How do you fit interdimensional monsters into a story that takes place during the months when the door is sealed? Easy: Particles from the Other Side that escaped before the door closed have begun to mutate into different plant species in Hawkins, generating hybrid creatures like a “snow shark.” All of this comes from the original visual style of the illustrator Meybis Ruiz Cruz and which, through animation, Netflix has wanted to bring closer to series from the eighties such as ‘He-Man’ or ‘The Real Ghostbusters’, but without losing sight of more recent proposals such as the animated films of the Spider-verse or ‘Arcane’. ‘Stories of 85’ obeys the tactic of keeping a franchise alive once it is impossible to do so with the original actors, through formats that alleviate technical and distribution demands. If this experiment works, we will undoubtedly see similar attempts with series like ‘Wednesday’ or ‘The Bridgertons’, and that without leaving Netflix. And, of course, if it works it will also continue on this side: according to the new showrunnerEric Robles, there are ideas to cover many other intervals in the official series. You have to milk it from somewhere. In Xataka | 16 premieres on Netflix: this week, the new ‘Stranger Things’, a rare British series and the return of Charlize Theron

Who is Johny Srouji and why this great unknown has just become the second most powerful person at Apple

by

For those who have been following Apple for a long time, Johny Srouji is no stranger. For the rest of the world yes, but after the appointment of John Ternus as CEO of Applethis Israeli engineer has become the second most powerful person in the company. The question is obvious: who is Johny Srouji? Who is Srouji and why does he matter?. Born in Haifa, Israel, in 1964 to a middle-class Christian Arab family, Srouji studied computer science at the Israel Institute of Technology (Technion) and graduated Summa Cum Laude in both his engineering and master’s degrees. He worked at Intel and IBM before Apple hired him in 2008 with a very clear assignment: to design the company’s first chips. He did much more than that. The revolution made chip. That first chip designed by Srouji was the Apple A4, which debuted in 2010 in the original iPad and the iPhone 4. From there, Srouji forged one of the most prestigious hardware careers in the recent history of the technology industry. The A7 of 2013 was the first SoC in using 64-bit architecture, and then there would come the revolution of the Apple M1 with which the company definitively got rid of dependence on Intel in its Macs. But his work goes beyond. His official title until now was senior vice president of hardware technologies, but it did not reflect the real scope of his work. Srouji not only led the chip design. Also that of batteries, cameras, storage controllers, sensors, displays, cellular modems and other critical components of the entire family of Apple devices. Almost everything that makes these products work the way they do is largely due to the work of Srouji and his team. With the new position, his responsibility expands and he will now control the entire cycle: not only the hardware itself, but also the physical design. It’s a colossal challenge, but if anyone seems prepared to take it on, it’s Srouji. He was about to leave. In December 2025 Bloombeg reported that Srouji had informed Tim Cook that he was seriously considering leave Apple in the near future. Two days later, Srouji himself published a message to his team denying the newsbut the damage was done. For Apple to lose Srouji would have been a disaster, and it is very likely that this new position is in part Apple’s response to that alarm signal. Textbook talent retention, but raised to maximum power. New position, new structure. In it internal communication that Srouji has sent to team employees, the engineer detailed how he will organize the division into five areas: Hardware engineering: led by Tom Marieb, an Intel veteran who joined Apple in 2019. Siilicio: it will be directed by Sri Santhanam, a manager with a long career at Apple Advanced Technologies: Supervised by Zongjian Chen Platform architecture: led by Tim Millet Program management: will be managed by Donny Nordhues In that message, Srouji acknowledges that this “represents a significant change” but believes it will work thanks to the entire team. It seems that you are very clear about how you want to work with your team. A fusion with a lot of historical sense. The reunification of hardware engineering and the hardware technologies division under the same leader is not entirely new. It is the structure that Apple had for years under the direction of Bob Mansfield, former head of hardware. until 2013 and? then he took charge of the failed Project TitanApple’s car. That’s when those two areas were divided, something that allowed both Ternus and Srouji to progress in their domains, but also caused some structural tensions between teams that had to collaborate. Bringing them back together is a clear commitment to strengthening that collaboration. The great cover-up of Ternus’s appointment. It is normal that the vast majority of headlines go to Ternus, who will decide the future of the company from now on, but Apple is above all a hardware company. That Srouji now becomes his leader makes this engineer a person with enormous power within the company. The change is promising in terms of promoting that facet of the product that both he and Ternus dominate, and without a doubt interesting times await us at Apple. Image | Apple In Xataka | John Ternus, vice president of Apple: “The iPhone Air had been in development for years, but we had to say ‘no’ until now”

Anthropic says Claude Mythos is too powerful to go public. The question is if this is nothing more than “the wolf is coming”

by

Claude Mythos Preview It is the best AI model ever created. We don’t say it, Anthropic says it, but almost no one else can say it because only a select group of companies has access to said model. The cybersecurity capabilities of the model appear to be astonishingbut more and more experts say that although Mythos is better than its predecessors, it is not the revolutionary leap that Anthropic seems to propose. Is that way of launching the model just an effective way of creating hype? Beware the Anthropic speech. The well-known entrepreneur and analyst Gary Marcus recently gave three reasons why, according to him, the launch of Mythos is not as revolutionary as Anthropic wants us to see. I cited tweets from software engineers and cybersecurity experts who cast doubt on Anthropic’s claims. The company published a study on the capabilities of Claude Mythos Preview that seemed to make it an extraordinary tool for the field of cybersecurity, but at the same time it was so powerful that it could be very dangerous if it fell into the wrong hands. Isn’t that a big deal? Among Claude Mythos’ achievements, Anthropic highlighted how he had found vulnerabilities in Firefox 147. But in reality many of the flaws were basically variations of the same two bugs. If you removed them from the equation, Mythos’ effectiveness rate at finding new exploits dropped a lot, even below Opus 4.6. Anthropic did not hide that fact, of course, but it makes this capacity, for example, not seem so striking. An X user also criticized the use of Cybench as a cybersecurity benchmark when Opus 4.6 almost completely surpassed it. For him, the choice of some of the Anthropic tests was debatable because they were not a challenge to current models. Other models can do the same. The co-founder and CEO of Hugging Face, Clement Delangue, stated that Mythos was no big deal. Their argument: they had used small, cheap open models, isolated the relevant code from some examples of the vulnerabilities found by Mythos, and they found the same problems which had already detected the Anthropic model. According to the Epoch Capabilities Index, which measures the capacity of AI models by combining several benchmarks, the leap that Mythos has taken is striking and “departs” from the progressive line of its predecessors. Source: Anthropic. Observer bias. But here it should be noted that in those analyzes they knew where to look because Mythos had already found those problems. We are dealing with observer bias, and in fact the Hugging Face document makes it clear that they even gave him specific clues such as “consider integer overflow”) to find those bugs. And on this observation, another one: Hugging Face does not say that a small model can replace Mythos on its own, but that it can be very good by giving it the appropriate code fragment. Mythos seems more capable of blindly complex security breaches, but it is a huge model and that is why it has greater capacity. Or what is the same: Mythos is better because it has the size, design and resources to be better. Fear, uncertainty, doubt? The language used by Anthropic in this advertisement could be considered to some extent a clear use of FUD (“Fear, Uncertainty, Doubt” -> “Fear, Uncertainty, Doubt”) as a marketing technique. It is a resource that has been seen in the past, and for example OpenAI already said in 2019—years before the launch of ChatGPT—that GPT-2 was too dangerous for a public launch. Obviously it wasn’t, but that certainly served to create expectation about the true capacity of the model. It’s better, but it may not be revolutionary. The results of the benchmarks that Anthropic published already made it clear that although there are very notable jumps in some tests, in others the evolution is much less striking. Claude Mythos was not the best at everything, and now analysts appear who contrast that data with other metrics. For example, with the Epoch Capabilities Index (ECI) from Epoch AI, the startup that has one of the most reputable benchmarks of the industry. And according to this index, Claude Mythos is above his rivals, but not for long. The wolf is coming. The truth is that the launch of Claude Mythos Preview has been really striking and the documents that accompanied that document tell us about a really capable AI model. The problem is that it is impossible to verify it because only a few companies have access to it and can test it. Without that public availability the only thing we can do is trust (or not) what Anthropic tells us, and that is the point: it is not clear that we should do it. The company is interested in us buying this discourse, obviously, but without an independent analysis it is impossible to verify these statements. In Xataka | Anthropic has become the darling of AI and has sought a partner to guarantee its future. It’s not the one we thought

Claude Mythos is an AI model so powerful it’s scary. So Anthropic has decided that you won’t be able to use it

by

Claude Mythos Preview it’s already here and it’s so good it’s scary. Literally. Anthropic has just introduced it to the public, but it has been done so cautiously that we won’t even be able to test it and it will only be available for certain technology partners. That’s frustrating and disturbing at the same time, but also reasonable. So powerful that it scares. On February 24, 2026, Anthropic engineers were able to test their new artificial intelligence model for the first time, which they called Claude Mythos Preview. As soon as they did they realized one thing: “demonstrated a dramatic leap in its cyber capabilities over previous models, including the ability to autonomously discover and exploit vulnerabilities zero-day in the main operating systems and web browsers on the market. Threat to global cybersecurity. This finding made it clear to Anthropic officials that although this capability makes it very valuable for defensive purposes, it also poses clear risks if the model were offered globally. Thus, a cybercriminal could take advantage of it to find vulnerabilities in all types of systems and exploit them. A few hours ago the company developed this analysis of Mythos as a threat to cybersecurity in a post on his blogand for example highlighted how Mythos found a vulnerability (now corrected) that had been present in OpenBSD for 27 years, an operating system precisely recognized for its very strong security. There were more examples, and all of them made the conclusion clear: Mythos is too powerful for ordinary mortals to use. Superior in all benchmarks, and in some cases such as USAMO (mathematics), the jump is simply incredible. Source: Anthropic. The best in history according to benchmarks. Anthropic has published a very in-depth report about this model with its “system card”. Among the data present is, for example, its performance in benchmarks, where it has swept GPT 5.4, Gemini 3.1 Pro and also Claude Ous 4.6, which until now was the best model in the world in almost all performance tests. Although in some cases the jump is not spectacular, in others such as USAMO —mathematical problem solving—Mythos practically achieves perfection. He barely hallucinates… That system card also talks in detail about how Claude Mythos Preview has a drastically lower hallucination rate than Claude Opus 4.6 and earlier models. He is also capable of saying “I don’t know” if he does not have enough information to answer, something that reduces hallucinations due to overconfidence. …but when it does, be careful. The paper warns of a new phenomenon: when the model fails in some complex tasks, the “hallucinations” are not obvious errors, but rather extremely subtle and well-argued technical failures. This is dangerous because the answer seems totally correct to experts, requiring very deep verification. Glasswing Project. That power and capacity has meant that the model will only be available through a “defensive” program that they have called Glasswing Project and which will be exclusive to some Anthropic technology partners. Specifically AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA and Palo Alto Networks. All of them will have the privilege (and responsibility) of having access to Claude Mythos Preview to identify vulnerabilities and exploits and correct them before bad actors can do so. Mythos Preview “it’s just the beginning”. Although this model is the most capable that has been seen so far, at least according to the benchmarks and data presented by Anthropic, the company assures that “we see no reason to think that Mythos Preview is the point at which the cybersecurity capabilities of language models reach their peak.” They assure that they expect the models to continue improving in the coming months and years, although this new model is certainly on another level. In Xataka | OpenAI and Anthropic have proposed the impossible: lose $85 billion in one year and survive

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.