Programming is the new board of AI. OpenAI and Anthropic have made it clear with GPT-5.3-Codex and Claude Opus 4.6

When ChatGPT broke out in November 2022, OpenAI seemed unrivaled. And, to a large extent, that was the case. That chatbot, despite its errors and limitations, inaugurated a category of its own. However, in the technology sector advantages are rarely permanent and, in 2026, the position of the company led by Sam Altman It’s a far cry from what it had then. Google has managed to attract the general public with Nano Banana Prowhile Gemini steadily gaining ground as an artificial intelligence chatbot. At the same time, ChatGPT’s market share has fallen significantly in some markets. Anthropic, for its part, has established itself as a reference in software engineering and has become one of the preferred tools among programmers. In this race to set the pace of AI, this Thursday we witnessed a curious movement: the almost simultaneous arrival of two models focused on programming, GPT-5.3-Codex and Claude Opus 4.6. The coincidence does not seem coincidental and reflects the extent to which the major players in the sector compete to define the next step, in a scenario where the main beneficiaries are, once again, the users. With these new models already on the table, the question becomes what they really contribute. There are plenty of promises and they are also beginning to appear benchmarks comparable that help to place them. So, therefore, it is time to look in a little more detail at what OpenAI and Anthropic propose for those who use AI as a development tool. GPT-5.3-Codex and Opus 4.6 enter the scene: what each promises to developers GPT-5.3-Codex is presented as a model focused on scheduling agents which seeks to expand the scope of what a developer can delegate to AI. OpenAI claims that it combines improvements in code performance, reasoning and professional knowledge over previous generations and is 25% faster. With this balance, the system is oriented to prolonged tasks that involve research, use of tools and complex execution, while also maintaining the possibility of intervening and guiding the process in real time without losing the work thread. One of the most striking elements that OpenAI highlights in this generation is the role that Codex itself would have had in its development. The team used early versions of the model to debug training, manage deployment, and analyze test and evaluation results, an approach that accelerated research and engineering cycles. Beyond that internal process, GPT-5.3-Codex also shows progress in practical tasks such as the autonomous creation of web applications and games. The company has published two examples that we can try right now by clicking on the links: a racing game with eight maps and a diving game to explore reefs. Anthropic’s turn comes with Claude Opus 4.6, an update that the company presents as a direct improvement in planning, autonomy and reliability within large code bases. The model, they claim, can sustain agentic tasks for longer, reviewing and debugging its own work more accurately. The idea is that we can use these capabilities in tasks such as financial analysis, documentary research or creating presentations. Added to this is a context window of up to one million tokens in beta phase, a leap that seeks to reduce the loss of information in long processes and reinforce the usefulness of the system. Beyond the core of the model, Anthropic accompanies Opus 4.6 with a series of changes aimed at prolonging its usefulness in real workflows. Among them there are mechanisms such as the so-called “adaptive thinking”, which allows the system automatically adjust the depth of your reasoning depending on the context. Configurable effort levels and context compression techniques designed to sustain long conversations and tasks without exhausting the available limits also appear on the scene. Added to this are teams of agents that can be coordinated in parallel within Claude Code and deeper Excel or PowerPoint integration. While OpenAI’s product, GPT-5.3-Codex, is not yet available in the API, Anthropic’s is. Maintains the base price of $5 per million entry tokens and $25 per million exit tokenswith nuances such as a premium cost when the prompts exceed 200,000 tokens. Measure who wins with numbers? When trying to put GPT-5.3-Codex and Claude Opus 4.6 face to face, the main obstacle is not the lack of figures, but rather their difficult correspondence. Each company selects evaluations that best reflect its progress and, although many belong to similar categories, they differ in methodology, versions or metrics, which prevents a direct reading. In this type of models, this fragmentation of results is part of the state of the technology itself, but also requires cautious interpretation that separates technical demonstrations from truly equivalent comparisons. Only from this filter is it possible to identify the few points where both systems can be measured under comparable conditions and draw useful conclusions for developers. If we restrict the analysis to truly comparable metrics, the common ground between GPT-5.3-Codex and Claude Opus 4.6 is limited to two specific evaluations identified through our own research: Terminal-Bench 2.0 and OS World in its verified version. The results show a distribution of strengths rather than a clear supremacy. GPT-5.3-Codex marks a 77.3% in Terminal-Bench 2.0 compared to 65.4% for Opus 4.6, which points to greater efficiency in terminal-centric workflows. On the contrary, Opus 4.6 reaches a 72.7% on OSWorldsurpassing the 64.7% of GPT-5.3-Codex in general interaction tasks with the system, a contrast that reinforces the idea of ​​specialization according to the environment of use. So we could say that the capabilities described by each manufacturer point to tools that are no longer limited to generating code, but rather seek to participate in prolonged processes of analysis, execution and review within real professional environments. This transition introduces new selection criteria that go beyond punctual performance. In Xataka | OpenAI has a problem: Anthropic is succeeding right where the most money is at stake

We have been avoiding aged cheese for years for health reasons. Massive study suggests we were wrong

For decades, nutritional guides and specific diets focused on ensuring brain health, such as the famous MIND diethave had a common enemy: saturated fats of dairy origin. However, science has now given a turn of the wheel to show us that we were completely wrong. New evidence. A new and comprehensive study published in the magazine Neurology You just turned this belief upside down. After following almost 28,000 people for a quarter of a century, researchers at Lund University have found a surprising association: regular consumption of high-fat cheese and cream not only does not increase the risk of dementia, but seems to reduce it significantly. The Swedish diet. The researchers conducted a median follow-up of 25 years until 2020, cross-referencing dietary data with the Swedish National Patient Registry. The result was that during this type 3,208 were identified cases of dementiaand from here we began to see what these people ate. In this case, those who consumed 50 grams or more of high-fat cheese per day showed a reduced risk of dementia of between 13% and 19% compared to those who did not consume it. Furthermore, consumption of high-fat cream was associated with a 16% reduction in the risk of having full-blown dementia. But there is more. The most curious thing about the finding was the specificity, since similar benefits were not found in low-fat dairy products, nor in regular milk or butter. In this way, you can see that there is something specific in the nutritional matrix of cheese and fermented cream that plays in favor of our brain. Why this cheese. Emily Sonestedt, co-author of the study, She was surprised by the resultsalthough he points out that they have biological logic. While traditional diets limit cheese due to its calorie and saturated fat content, this food is rich in medium chain fatty acids, vitamin K2calcium and high quality proteins. In addition to all this, the fact that it is a fermented food can positively influence the intestinal microbiota, and we know more and more about the direct connection between the intestine and the brain. In this way, maintaining a good microbiota again indicates that it guarantees us having better brain health. You have to be cautious. Before running to the supermarket to buy all the types of cheese on the shelves, it is necessary to put on the usual handbrake in science, since we are talking about an observational study. This means that science points out that two things happen at the same time, but it does not prove 100% that one causes the other. And in this case, lifestyle may be interfering, such as the fact that people who eat cheese in Sweden have other lifestyle habits such as greater physical activity that protect them, although the researchers tried to adjust the variables. The verdict. The idea that “all saturated fat is bad for the brain” is losing steam in the face of evidence that certain complex foods, such as aged cheese or cream, have properties that go beyond their basic nutritional label. As is often the case in nutrition, the key does not seem to be eliminating food groups, but rather understanding the quality and source of what we eat. Images | Aliona Gumeniuk Robina Weermeijer In Xataka | Forgetting things is not a bug, it is a feature of your brain: how not remembering things makes us think better

Mexico was supposed to be giving oil to Cuba out of “humanity.” Now we know that he was charging millions

On the coast of Veracruz, Mexico’s diplomatic and energy machinery has applied the handbrake. The image of the ship Ocean Marinerdocking in Havana on January 9 with 85,000 barrels of crude oil, seems to be the last postcard of an era that is abruptly closing. As confirmed France 24that was the last successful shipment before geopolitics cut off the flow. His replacement, Swift Galaxywas scheduled to sail in mid-January, but his trip was quietly canceled and he disappeared from the logistical calendar of Mexican Petroleum, how they have advanced in The Country. What happens in Mexican ports is the reflection of a tension that goes beyond commercial matters. After the American intervention in Venezuela on January 3 and the fall of Nicolás Maduro, the president of the United States, Donald Trump, was blunt: “No more money or oil will reach Cuba. Zero.” The threat was accompanied by an executive order that promises tariffs on any nation that supplies crude oil to the island, which Trump has described as a “failed nation.” Caught in this crossfire, Claudia Sheinbaum’s government navigates between two waters. On the one hand, it defends the “sovereignty” of helping a sister nation; On the other hand, in the Washington offices, their own accounting books tell another story: formal businesses and punctual payments that refute the purely humanitarian narrative. Solidarity after the storm From the National Palace, the speech has tried to avoid direct confrontation appealing to history. President Sheinbaum has reiterated that Mexico, faithful to its diplomatic tradition of voting against the blockade from day one, has the sovereign power to decide whether to “sell or give” oil to Cuba. This rhetoric gained strength at the end of 2024. After the collapse of the Cuban electrical system and the devastating passage of Hurricane Rafael in November, the Mexican government started labeling their shipments under the umbrella of “humanitarian aid.” However, here the enigma arises. Although the president assures that there is a humanitarian donation channel other than the commercial one, her administration has not offered specific figures on how many barrels are given away and how many are charged. Everything is opacity in the help, while the business has lights and stenographers, as highlighted The Country. While the political discourse focuses on solidarity, the financial documents are cold and exact. Pemex, which is listed on international markets, cannot afford ambiguities before the United States Securities and Exchange Commission (SEC). According to the information delivered to this regulatory body, the Mexican oil company maintains a current contract with the Cuban government since July 2023 through its subsidiary Wellbeing Gasoline. Far from being a hidden charity, the figures revealed by the director of Pemex, Víctor Rodríguez Padilla, show an active and lucrative commercial relationship. In 2025, Mexico sold oil to Cuba worth 496 million dollars. If we add what has been invoiced since the start of the contract in 2023, the total figure amounts to about 1.4 billion dollars. Rodríguez Padilla was emphatic in denying that Cuba does not pay its debts, a common perception given the island’s crisis. “Of course they pay us! We have a business relationship too. They are very formal in their payments,” the manager assuredclarifying that there are no overdue invoices. To try to minimize the impact of these revelations before the scrutinizing eyes of Washington, Pemex has argued thatAlthough the figures sound high, they are marginal for the company: they represent less than 1% of its crude oil production and just 0.1% of its oil sales. It is an “open” contract that depends on Mexico’s availability, and not an unbreakable commitment. The domino effect: why the tap was turned off The current crisis is not explained only by Mexico’s decisions, but by the collapse of Havana’s historical suppliers. For years, Venezuela was the island’s lifeline, shipping up to 100,000 barrels a day during the time of Hugo Chávez. However, after the capture of Nicolás Maduro and the US intervention in Caracas, these shipments ceased completely in January. as detailed BBC. Mexico then became the last lifeline, sending approximately 20,000 barrels a day, a figure that, although far from the island’s total needs, was essential. to maintain minimum services. The pressure escalated when Republican congressmen, such as Carlos Giménez, put the Treaty between Mexico, the United States and Canada (T-MEC) on the table. The threat it was clear: If Mexico continues to oxygenate the Cuban regime, the review of the trade agreement in 2026 could become a nightmare for the Mexican economy. Faced with the risk of tariffs that would damage its own economy, Mexico chose to suspend hydrocarbon shipments. The consequences of this supply cut are immediate and alarming. A graph made with data from Kpler and published by the Financial Times illustrates the seriousness of the moment: Cuba’s crude oil imports have plummeted and, according to the estimates displayed in the report, the island only has oil reserves left for between 15 and 20 days. The situation has raised alarm bells at the United Nations. The Secretary General, Antonio Guterres, he warned through his spokesperson that Cuba is at risk of imminent “humanitarian collapse” if its energy needs are not met. Without fuel, not only do the lights go out; The pumping of drinking water, the transportation of food and the operation of hospitals are stopped. Faced with the impossibility of shipping oil without suffering commercial reprisals, the Sheinbaum government has modified its relief strategy. The president confirmed that, while the Foreign Ministry seeks “diplomatic ways” to resolve the oil issue, Mexico will ship this week shipments of food and basic products managed by the Secretary of the Navy. It is a palliative for a crisis that is, above all, energy. In this maximum pressure scenario, an unexpected edge arises. As Trump closes the oil fence, he has also dropped comments that suggest the door is not completely closed. The American president recently stated that “we are negotiating with Cuban leaders right now,” hinting at conversations about immigration issues and the … Read more

Google has smelled blood with AI, so it has decided to spend more in 2026 than the GDP of 158 countries in the world

New year, new budgets. Big tech companies are beginning to detail their roadmap for 2026 and the trend is clear: spend even more on AI. a few days ago, Goal announced that the planned capex (capital expenditure) rose to 135,000 million dollars and Microsoft too pointed to a similar figure. Alphabet (Google) just told everyone to “hold my hands.” May the rhythm not stop. The bomb was announced during the last results conference. Alphabet plans to spend between $175 and $185 billion, doubling 2025 capex, which was $91.4 billion, and almost quadrupling 2024 spending (52.5 billion). To put it in context, it is more than the GDP of Morocco, Kuwait, Bulgaria and up to 158 countries. At the same time, the company announced record results, surpassing 400 billion in revenue for the first time. The net profit stood at 132,000 million. Vertigo. That’s what investors seem to have felt. They count in Financial Times that, in the hours following the news, Alphabet shares fell 7% after the capex announcement, but then the fall was reduced to -1.5%. Microsoft experienced a similar response after its earnings call a few days ago, it is the response of investors to these exorbitant figures. However, as long as the results are good, it seems that the scare will not last long. Everything’s fine. They count in Fortune that Pichai assured that this year’s capital expenditure is “a look at the future” and justified his strategy by highlighting that the demand for his cloud services and DeepMind (Gemini) is extraordinary, so the investment must also be. He also announced that AI searches now surpass traditional searches and that Google Search’s business has grown 17% compared to last year. Additionally, the order book for its cloud has increased by 55% during the last quarter. It still won’t be enough. The CEO of Alphabet admitted that, despite the record results, there are insurmountable bottlenecks such as computing capacity, problems in the chip supply chain and energy limitations. These restrictions make it take a long time to get a data center up and running, or in other words, it was preparing investors not to expect an immediate return. Gemini, full out. The Google chatbot is in its sweet moment. The viral success of Nano Banana, Gemini 3 sweeping its competition in benchmarks and Apple choosing him as the new brain for the new Siri They have given a boost in popularity to Gemini, which already has more than 750 million users. OpenAI is still ahead with ChatGPT, but Google is closing the gap and Altman’s people have reacted going into panic mode. He moat of Gemini. Benchmarks are fine, but there is something much more important. During the conference, Pichai announced that they had reduced Gemini’s service costs by 78% “through model optimizations, efficiency and utilization improvements.” It is no longer that its AI is surpassing its competition, it is that it is cheaper and there OpenAI does have a problem. With its advertising businesses, the cloud and more revenue, Google has plenty of room to skyrocket its capex. In Xataka | OpenAI’s entire financial strategy depended on achieving a monopoly with ChatGPT: the opposite is happening Image | Wikipedia

Door-to-door services to skip the MOT

A Civil Guard arresting a Civil Guard. He and five other people have been brought to justice in what has been an operation to dismantle a network of fraud related to the ITV exams. This is what they point out in EFEwho have echoed what happened in Ondara (Valencia). They explain in the news agency that the Civil Guard has arrested a total of six people who were dedicated to helping their clients pass the ITV exams when they were going to receive an unfavorable result. The station workers, who have been suspended from employment and salary for the moment, would be collaborating in the plot, and a Civil Guard whose position is still unknown. According to media outlets such as The Provincesamong those arrested and investigated there are also regular customers of the ITV station who would have been benefiting from the blind eye that the workers would have turned. These workers are, for the moment, suspended from employment and salary, according to EFE. At the moment, it is believed that the Civil Guard was aware of the operations and collaborated with the workers so that the drivers passed the ITV, in theory unfavorable, on their vehicles. However, in the diary Levantpoint out that the work of the Civil Guard would go even further. They point out that he could have been receiving commissions in exchange for workers looking the other way and, on some occasions, he would even have offered a kind of “door-to-door” service, taking the car himself to the technical inspection station and returning it to its owner once he received the go-ahead. This newspaper points out that the agent was discovered shortly before Christmas when his name appeared in another case. Since then the Civil Guard would have their conversations intervened, which has been key to dismantling the plot. What happens if I don’t pass the ITV? Do not pass the MOT because the car has serious defects It’s a mess. It’s a real big mess. Because a car that has been given a “negative” result in this mandatory test can only leave the station on a tow truck. In that case, the car must go directly to the workshop, without touching the road, and return in the same way again to the ITV station. Something more or less similar happens when we receive an “unfavorable” result on the exam. In this case, the circulation of the vehicle is also prohibited except for the route from the ITV to the workshop and from the workshop to the ITV. In that case, the driver can take his car on the road. Failure to comply with the requirements has such serious consequences that many drivers skip the exam. And the data says that more cars skip it the older the vehicle is because the maintenance to keep the car up to date on, for example, the emissions test may not compensate if we take the value of the vehicle as a reference. It must be taken into account that a good part of these vehicles that stop attending the ITV are work vans who have to pass their exams every six months or mopeds low price. So much so that it is estimated that 25% of the motorcycles circulating in Spain do not comply with the ITV. The problem is that fines are sometimes cheaper than fixing the damage on an old vehicle. And the penalties for driving a vehicle that does not comply with the ITV approval are the following: Circulating without ITV: 200 euros, 100 euros with prompt payment. Circulate with unfavorable ITV: 200 euros, 100 euros with prompt payment. Driving with a negative ITV: 500 euros and immobilization of the vehicle, 250 euros prompt payment. Photo | AECA-ITV In Xataka | Documentation to pass the ITV: everything we have to bring

While Europe studies reintroducing military service, Mexico has taken the opposite path: reducing it

The Russian invasion of Ukraine and the shift in international policies of the US have caused Europe to no longer trust NATO as a defensive shield, betting on improve your defense resources. Thus, while several European countries debate whether to return to introduce military service mandatory, Mexico decides to take the opposite path and shorten the mandatory military training of its citizens so that it fits better into the lives of young people and is more attractive to them. The change of Mexico. The Government of President Sheinbaum has applied the largest operational change in the conditions of the National Military Service (SMN) in Mexico since 1942. As stated in the article 5 of the Political Constitution of the Mexican States, the service of arms is mandatory for all Mexicans between 18 and 40 years old. This call-up is divided into two modalities: Framed and Available. The former remain quartered for about three months, while the latter remain at the disposal of the Secretariat of National Defense (SEDENA) for a year, during which time they are instructed in training sessions on Saturdays. However, with the last reform which has come into effect in January 2026, the training phase has gone from 44 weeks to just 13, with limited classes for those assigned as “On Availability” on Saturdays from 7:00 a.m. to 1:00 p.m. so that the fulfillment of this duty does not interrupt the studies or work of recruits. New civil-military program. Colonel Juan Sandoval Muñoz, commander of the 78th Infantry Battalion, explained to The Universal that the new “Availability” model prioritizes civil-military subjects to reinforce military values ​​and discipline. The dates for this training are divided into two periods: from February 14 to May 9, and from August 1 to October 24, adding a total of 13 sessions. Participants cover 10 subjects focused on basic training, discipline and support for the population, first aid, civil protection and DN-III-E Plan and knowledge of weapons. All subjects are taught by officers and sergeants of the Armed Forces so that recruits become familiar with the military hierarchy. 13 weeks in the barracks. For its part, the Framed option remains with 13 weeks of admission to barracks, with tailored training as if they were professional soldiers. According to Colonel Sandobal, many of these recruits requested this modality to release their SMN Identity Card in less time, which certifies that they have fulfilled their duty, but they ended up requesting entry into the army or military schools. For this reason, it was decided not to change this modality to keep this recruitment route open, despite being equal in time to the other alternative. European rearmament brings the military back. In Europecountries like Denmark accelerate compulsory military service from 2026 for womenwhich was previously only voluntary, and they extend it to eleven months in its basic version. For its part, Germany is discussing bringing back the voluntary military service before the end of the year, after abandoning it in 2011, and Croatia reactivates it on a mandatory basis with a duration of two months for men starting in January 2026. Other countries like france and Poland are starting ten-month voluntary programs for 18- to 19-year-olds, with the option of joining the reserve or the army if a military crisis arises. In Spain the Minister of Defense continue betting by a voluntary reservation instead of resume military servicebut that does not prevent the debate circulate in the army. At least on a theoretical level. In Xataka | In the midst of rearmament, Europe has realized an unimportant detail: it does not have enough bullets Image | Government of Mexico, Unsplash (JEsus Herrera)

that parking your car costs more than a ticket

We have been talking about for years the drift schizophrenic that has been generated around large live events, and especially music concerts. The phenomenon has led us to accept that if you want to go to a Bad Bunny concert or Radiohead The formula that never fails is to multiply the original price by three and go to the slaughterhouse from resale. Now, someone unexpected has wanted to go further by opening a parallel market to one of the most anticipated events. The World Cup car parks… and their resale, of course. Pay before entering. The 2026 World Cup to be held in the United States and Mexico is shaping up to be the tournament most expensive ever organized in history, and not only because of the price of the tickets, which are already supposed to be astronomical, but because of everything that surrounds the simple act of going to the stadium. In an urban context designed for the car and not for the pedestrian such as the United States, the fan’s experience begins long before access control and becomes a sum of tolls that raise the real cost to unprecedented levels in the history of football. Parking as a core business. The ironic thing, or perhaps not so ironic, is that it is FIFA itself that has started selling parking passes for figures that range between 75 and 175 dollars in minor games, but that in key venues like Los Angeles reach, attention, nothing less than up to 250 and 300 dollars per vehicle and match… even when those spaces are located more than a kilometer from the stadium. In practice, parking already costs the same (or more) than many official tickets, so a round of 16 or quarterfinal match with a ticket of 400 or 500 dollars and a parking lot of 300 easily raises the total bill per person to around $1,000, a figure that redefines what it means to “go to football.” Stadiums far away, car mandatory. It happens that, unlike in Europe, where large stadiums are usually integrated into the city and connected by subway, train or bus, many World Cup venues in the United States are located in peripheral areas and were specifically conceived to arrive by car. This structural dependency turns the parking lot into an essential resource and allows it to be monetized as part of the show, something unthinkable in most European venues, where even when some clubs already charge for parking (as happens at Atlético de Madrid) there is always the real and massive alternative of public transport. Planned shortages and inflated prices. Plus: the problem is not only the price, but deliberate scarcity. Many parking spaces near the stadiums will remain within security perimeters or reserved for sponsors, drastically reducing the offer for the general public. In cities accustomed to tens of thousands of seats at NFL events, the World Cup will put only a fraction on sale, creating a bottleneck perfect to justify exorbitant prices under the argument of the “local market” and “large comparable events”. Hello resale. Yes, this brings us to an “old acquaintance” of any massive event worth its salt: resale. They counted this week in The Athletic that, as has happened with the concert ticketsparking has fully entered the speculation circuit, with passes resold in secondary markets for even higher figures, although it may seem difficult. In fact, this occurs even in venues where FIFA has not yet published your final offer. The result is a general feeling of abuse, in which the fan pays not only to watch the game, but for each step necessary to get there. A deja vu. This escalation is not an isolated phenomenon and we have been counting in the last few yearsdocumenting how concerts and big events live have entered a price spiral marked by dynamic rates, uncontrollable resales and added charges that turn the luxury experience. The 2026 Soccer World Cup adds to that logic: tickets that are difficult to get at the official price, crazy resale and peripheral costs (such as parking) that equal or exceed the show itself. Parking as a symbol of a new frontier. The underlying message is crystal clear and deeply uncomfortable: the World Cup, a phenomenon of masses and global audiences, is going to explode. a new business without any shame, that of making parking the car cost the same (or even more) than a ticket to watch the games. It is not a trivial detail, nor logistical, nor even collateral damage, but one more piece of the economic model of the tournament, designed to maximize income in each phase of the fan’s journey. The curious thing is that football remains the same as always on the pitch, but off it, getting to the stadium has become part of the bill. If we are not in hell, the truth is that we are getting closer with the tips of our fingers. Image | Ron ReiringRawPixel In Xataka | RTVE wants to win the ratings war at any price. Although that price is 55 million for the World Cup In Xataka | The Government believes it has the solution to the very serious increase in ticket resale prices. It may just make it worse.

democratize the climb to Everest

a drone DJI Mavic 3 Pro Equipped with a Hasselblad camera, he has managed to capture something that until recently seemed impossible: the complete ascent of Mount Everest from base camp to the summit in a single continuous flight. For 43 minutes, the aircraft traveled 3,500 meters of altitude, crossing the Khumbu icefall, the South Col and the final walls until reaching 8,848 meters of altitude. The images reveal the normal route of ascent in all its magnitude, including the characteristic queues of mountaineers who each season try to crown the roof of the world. The challenge. At that altitude, the air contains barely a third of the oxygen available at sea level, temperatures can drop to -30°C and winds reach speeds that would make the flight of any conventional drone impossible. The team used the Mavic 3 Pro with a four-thirds CMOS sensor, a combination that allowed stability and image quality to be maintained in extreme conditions. Beyond the visual spectacle, this flight is part of a more ambitious project by DJI: to demonstrate that drones can save lives on the highest mountain on the planet. Drones in high mountains. The tests of DJI on Everest respond to a clear commercial strategy: convert their drones into rescue and logistics tools in extreme environments. The Chinese company seeks to demonstrate that these aircraft can transport medicines, locate missing mountaineers and facilitate emergency operations at altitudes where thin air complicates any human intervention. The best-known precedent occurred in 2018, when Scottish mountaineer Rick Allen was located on Broad Peak after 36 hours lost at more than 7,000 meters of altitude. thanks to a DJI Mavic drone. That rescue, coordinated by Bartek Bargiel, brother of the skier Andrzej Bargiel, who we will now talk about again, marked a turning point in the perception of drones as high mountain safety instruments. Qualitative leap. In 2025, the Nepalese company Airlift Technology began providing drone logistics services between Everest Base Camp and Camp One, separated by approximately 2.9 kilometers in a straight line but by a 700-meter difference in altitude and the dangerous Khumbu Icefall. What takes the Sherpas between six and seven hours of crossing, a drone completes it in six or seven minutes. Milan Pandey, the company’s drone pilot, explains that during the 2025 climbing season they transported ladders, ropes and oxygen cylinders following radio instructions from the Sherpas who install the fixed routes. Safer. The impact on the job security of these high mountain workers is significant. The so-called “icefall doctors” (Sherpas specialized in preparing and maintaining the passage through the Khumbu glacier) traditionally had to go up and down dozens of times each season carrying heavy equipment through unstable terrain where Almost 50 people have died since 1953. They can now request additional material without having to descend to base camp, which dramatically reduces risk. The key case. On September 22, 2025, Polish mountain skier Andrzej Bargiel completed a feat that combines extreme mountaineering with technological innovation: Ascended Everest without supplemental oxygen and skied down to base camp without removing skis. After almost 16 hours climbing in the so-called “death zone” above 8,000 meters, Bargiel began the descent along the South Col route. What was innovative was the role of drones in this expedition: his brother Bartek piloted one from base camp to guide him through the Khumbu Icefall, the most dangerous section of the descent. All this is seen in the full 31 minute documentary which records the adventure using cameras mounted on Bargiel’s helmet and aerial shots captured by drones. The footage reveals a extremely technical descent: ice, almost vertical walls, traversing exposed ledges and, in the final stretch of the Khumbu waterfall, slow maneuvers avoiding deep cracks and blocks of ice the size of buildings. The assistance of the drone was critical precisely in this sector: Bartek flew in real time over the glacier, identifying stable snow bridges, marking dead ends and choosing safe slopes. Visual democratization. Videos like these are part of a broader phenomenon. YouTube hosts thousands of recordings documenting mountain climbs, cave explorations, glacier traverses, and cliff flyovers that until a decade ago could only be captured by helicopters or million-dollar productions. An example is that of the Chinese photographer Ma Chunlin, who spent five years obtaining the necessary permits and carrying out test flights before achieving a definitive recording of the ascent of Everest. in one shot. Technically possible. This type of content responds to a technological evolution that has made accessible tools previously reserved for professionals. Models like the DJI Mavic Mini, which weighs 249 grams, allow users without prior experience to capture stabilized aerial shots in resolutions higher than Full HD. Portability is key: Foldable drones that fit in a backpack during long hikes have removed the logistical barriers that previously limited aerial photography to specialized equipment. The doubts. The proliferation of drones in natural spaces has generated debates about their impact. Regulations vary significantly between countries and regions: some National Parks prohibit its use entirely, while others allow flights with prior authorization. The balance between visual access to nature and the preservation of these environments (including the protection of wildlife that may be disturbed by the noise and presence of these devices) remains an open question. In Xataka | China doesn’t know what to do with so many drones. Their solution: create lower airspace

AI is very comfortable inventing everything it doesn’t know. Some researchers think they know how to stop him

The hallucinations have been the Achilles heel of AI since chatbots began to be part of our lives. Companies like OpenAI promised that hallucinations could be mitigated with adequate training processes, but years later both ChatGPT and its direct rivals They keep making up answers when they are not sure what to say. Shuhui Qu, a researcher at Stanford University, believes she has found a way to address the problem. A structural problem. Current language models have a factory defect: they respond with complete security even when they have no idea nor the necessary information. This has to do with how they progress when processing any answer, since LLMs have no problem completing the missing information, even if they are not being faithful to reality and are working with assumptions. First thing, recognize it. Shuhui Qu, a researcher at Stanford University, publishes an article in which she introduces what she calls Bidirectional Categorical Planning with Self-Consultation. An approach that starts from a simple idea, but uncomfortable for large technology companies: forcing the model to explicitly recognize what it does not know and not move forward until solving it. A more scientific method. The idea is not that the model think betterBut stop pretending you know everything. The approach of What starts from a basic premise: every time the model takes a step in its reasoning, it should ask itself if it really has the necessary information to do so. When an unknown condition appears, the model cannot continue. You are not allowed to fill the gap with an assumption, and you have to stop to resolve the uncertainty before moving forward. You can do this in two ways: Well asking a specific question to obtain the missing information Either by introducing some intermediate step (verification, additional consultation) that becomes part of the chain of reasoning. The method. The researchers, using external code, made models like GPT-4 They responded only when they had complete information. They did it with simple tasks, asking about cooking recipes and Wikihow guides. The key? They purposely withheld information to force him to stop. The conclusion of the research was that making preconditions explicit and verifying them before moving forward significantly reduces LLM errors when information is missing. Of course, along the way it is admitted that even this is not enough to make the hallucinations disappear completely. not so fast. Although the researcher’s idea sounds brilliant, it is quite unlikely to see it in the short and medium term. This way of processing breaks the natural flow of current LLMs, designed to return complete answers. To make such a system work, it is necessary to add an additional layer to the structure, some preconditions that force it to control the calls, interpret the responses themselves, classify them and self-block from asking questions if they do not have all the information. In other words, for the moment, AI will continue to score the triples to which we are already accustomed. Image | Xataka In Xataka | ChatGPT invents data and that is illegal in Europe. So an organization has set out to fix it with a lawsuit

investors are in “total caution” mode

In June 2025 everything was joy in the crypto world. one bitcoin reached the record value of $124,752 (according to CoinMarketCap) and marked a new historical record. From that moment, falls and more falls that have been done especially in recent days. And it’s not just bitcoin of course: it’s all cryptos. Bitcoin at $70,000. In the last 24 hours we have seen how bitcoin has barely managed to stay at the $70,000 barrier, and on some platforms it has even traded below that level. Right now it is around $71,600, but even with that data the conclusion is clear: in eight months bitcoin has lost more than 40% of its value. Risk aversion. We are seeing how the technology stock markets are falling quite generally in recent days because the results of the last quarter of the year have not been as good as expected. Even the gold, which was rising like foamhas also regressed. Investors are reducing risk overall, taking profits (and minimizing losses) and adopting much more cautious stances. In this scenario, BTC behaves as a risk asset, not as a safe haven, so divestment is the strategy that is being extended. what has happened. The macroeconomic situation is especially complex right now. Analyst Joe DiPasquale explained in Forbes how there is no internal problem in cryptocurrencies, but rather it is the global economic context that has caused the collapse: There are assets that are very sensitive to market movements: if the market rises a little, they rise a lot, but if it falls, they collapse. This is what we are seeing with cryptocurrencies, which are, in their opinion, these types of “high-beta” assets. Bitcoin and cryptos act as a kind of augmented “mirror” of how much money is left over in the economy. When there is a lot of money circulating (liquidity), bitcoin rises quickly, but if it is scarce, bitcoin is the first asset to be sold. The real economy weighs, and a lot. Government bonds are up and paying more interest right now, so investors prefer that security. The dollar also rose in value and strengthened, and since it is “more expensive”, you need fewer dollars to buy the same amount of bitcoin, which pushes the price down. But above all, as we said, investors have gone into total caution (“risk-off”) mode and have been selling volatile assets to protect their money in cash or gold. If the stock market falls, cryptos fall (more). CoinDesk also highlights that the 7.5% drop in bitcoin value In the last 24 hours it precisely followed large falls in Asian assets. There is concern about excessive spending on AI – fear of the bubble bursting again –, exaggerated valuations and lack of that increase in income that everyone promises. Google, Qualcomm and AMD—which fell a spectacular 17% yesterday, Wednesday—are some of the examples of these technological falls. Source: Alternative From greed to fear. The situation is very clear if we look at the Greed and Fear Index (Fear & Greed Index) from firms such as Alternative, which place it at a value of 12, or what is the same, “Extreme fear”. This index studies market movements and analysis to give that number, which is a great summary of the scenario we are experiencing. For much of 2025 that level was above 50 and reached 80 (extreme greed), but the current drop is evident. Bad time for cryptos. Of course, the collapse of bitcoin is as contagious as ever, and practically all cryptocurrencies have registered notable falls in recent weeks. Ethereum, which in August reached close to $4,800, is now in just 2,100. XRP has gone from 3.5 to 1.4 in that same period, and Solana from 247 to 91. Crypto believers are once again seeing their patience tested, but the maxim for them remains clear: HODL. In Xataka | In 2011 a group of investors bought 80,000 bitcoins. They just sold them 17,000,000% more expensive

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.