released

DeepSeek has just released a model that competes with Opus 4.6. It costs seven times less and runs on Chinese chips

by

They have passed 484 days since that “DeepSeek moment“, but the wait It seems to have been worth it, because we have the new DeepSeek V4 with us. We are facing an absolutely gigantic open weights model that once again promises to crack the foundations of the proprietary foundational models of Anthropic, OpenAI or Google. This is moving, gentlemen. Gigantic and open. DeepSeek v4 is an Open Source model and comes in two versions. The first is the Pro, with 1.6 trillion parameters (1.6T), of which it has 49,000 million active. The second is Flash, with 248,000 million parameters (248B, huge for a “Flash” model) of which 13,000 are active. More efficient than ever. Both versions they make use of a Mixture-of-Experts (MoE) architecture, which means that only a fraction of the parameters are activated in each inference. This allows the computational cost to be reduced significantly. Both versions support a context window of one million tokens—to include novels and novels at once as input—when in v3 it was 128,000 tokens. Furthermore, this model is much more efficient than its predecessor in computing per token: it requires only 27% of the operations per token and 10% of the KV cache compared to DeepSeek v3.2. Benchmarks promise. DeepSeek’s internal testing reveals that v4 Pro-Max (the best model with the highest reasoning ability) outperforms or is on par with Claude Opus 4.6 Max, GPT-5.4 xHigh, Gemini 3.1 Pro High, Kimi K2.6 and GLM 5.1. The results, however, are not independently verified, which means we should take them with caution. The numbers are still striking: in LiveCodeBench, a programming test, DeepSeek v4-Pro-Max achieves a 93.5% score compared to 88.8 for Opus 4.6 and 91.7% for Gemini 3.1 Pro. In other tests there is more variability, but at least on paper DeepSeek v4 Pro seems as good as Opus 4.7, which until now was the absolute benchmark. Much cheaper. But as happened with its previous version, the difference in price with those models from US companies is astonishing. As point the analyst Simon Willinson, the official prices of DeepSeek v4 Pro are 1.74 dollars per million input tokens and 3.48 dollars per million output tokens, up to almost seven times less than those of Opus 4.7 and up to almost 9 times less than those of the new GPT-5.5. With DeepSeek v4 Flash the cost is 0.14/0.28 dollars per million input/output tokens, when GPT-5.4 Mini costs up to 16 times more. The conclusion is obvious: if it really does what it says it does, the price is an absolute bargain. That is precisely the challenge: that real experience confirms what the benchmarks say. The hardware mystery. DeepSeek has not revealed what hardware has been used to train this version of its founding model. In the past they did admit that they had used NVIDIA’s H800s. Which yes it is known The thing is that the model has been developed to run on both NVIDIA and Huawei Ascend chips. This last has confirmed Baidu that its Ascend Supernode clusters based on the Ascend 950 will fully support DeepSeek v4 versions. Huawei support is “horrible” news for the US. In The Information they already commented that one of the reasons for the “delay” in the appearance of this model was to adapt it so that it worked without problems with Huawei chips. That support is according to Jensen Huang “horrible” news for the US, because it means that dependence on NVIDIA chips no longer exists or at least is reduced to a minimum. But. The launch comes at a difficult time for the company. Guo Daya, one of the people responsible for the v1 and v3 models, has signed for ByteDance to work on AI agents. Luo Fuli, who led the development of v2, joined Xiaomi last year. This launch also coincides with DeepSeek seeking external funding for the first time. They are expected to raise about $300 million and obtain a valuation of about $20 billion. according to The Wall Street Journal. From the surprise effect to the continuity effect. The launch of DeepSeek R1 in January 2025 was surprising because it demonstrated that China could train competitive models at a fraction of the cost of Western models. With DeepSeek v4 that surprise effect disappears to give way to the continuity effect. This model seems to maintain precisely what made the previous model famous: extraordinary power at a very low cost. Bad news for Anthropic. Such low prices are terrible news for Anthropic, which in recent weeks has been forced to execute a kind of “reduflation” of their new modelswhich are not more expensive but consume many more tokens. We’ll have to see if DeepSeek v4 Pro is as good as the company promises, but if it is, we’ll have another “DeepSeek moment” before us. Maybe not as notable as last year’s, but equally relevant. In Xataka | DeepSeek promised them happiness as the great Chinese AI. I didn’t count on a small detail: Kimi

In 2003 someone released 18 Bavarian beavers into the Ebro basin without saying anything. They have already arrived in Catalonia

by

It was a matter of time. In 2005 and while studying the European mink on the banks of the Aragón River, biologist Juan Carlos Ceña realized that something didn’t fit. There were felled trees, remains of forage, footprints, burrows and very specific droppings: it was just what one would expect to find in the vicinity of a beaver community. But there were no beavers in Spain. Everyone knew that. The strange story of the Iberian beavers. For years, researchers have debated whether the last specimens disappeared in the 17th century, the 18th century, or even the 19th century. In the end, the consensus is that the only evidence available They place them in the 2nd century BC. After that moment, no one knows what happened to the peninsula’s beavers. Therefore, what Ceña had just discovered was a bombshell. But, as soon as they started investigating it, they realized that there was a lot of fabric to cut. Sometime in the spring of 2003, someone illegally introduced 18 European beavers from Bavaria. Nobody knows for sure who he was or why he did it. But we know that it continued to be done. Today, there are beavers in the Tagus and the Guadalquivir. And of course we know that your beaver expansion it’s not natural. In 2023, biologist Teresa Calderón calculation that the Tormes beavers would have taken 40 years to get there by their own means from the closest documented population. The Andalusian case is more bloody: there is no way for the beavers to travel on their own the 365 kilometers of southern subplateau between the stretch of the Guadalquivir where they were found in 2023 and the closest point where we had previously found them. The ‘beaver bombing’ was a reality. But the worst was not (only) that: the worst was that, once they reached a river, they were there to stay. As soon as they took root in an area, they did not abandon it: if in 2007 they had already ‘conquered’ 60 kilometers of riverbank, by 2023 the beavers were already in Mequinenza and the lower stretch of the Ebro. It was a matter of time before they arrived in Catalonia and the news is that they have already arrived. The Center for Ecological Research and Forestry Applications has confirmed the presence of the beaver in the Segrià region, in the province of Lleida. Good news. And I’m not talking about the expansion of the beaver. That, today, is neither good nor bad news. It just is. I’m talking about, according to a handful of recent articles, “Beavers can turn river corridors into permanent carbon sinks“That is, they can be a climate ally that helps us recharge aquifers, purify water naturally and help in the recovery of wetlands. It is the ecological version of the old Castilian saying that “when God closes a door, he opens a window.” And thank goodness, because invasive species are here and we will not be able to get rid of them. Image | Derek Otway In Xataka | 20 years ago someone thought it was a good idea to release beavers into the Ebro. Now Zaragoza has a problem that is difficult to solve

When Sora was released many assumed it was “the death of Hollywood.” Only two years, then Sora no longer exists

by

In February 2024, OpenAI published on X a string of AI-generated videos with his new model, Sora. Although today, after two years of progress, they even feel outdated, at the time the result was convincing enough for the media around the world to start headlines that Hollywood had a very serious problem. Two years later, Sora does not exist. Panic effect. The effect of this presentation with videos was immediate: MIT Technology Review, for example, described them as “impressive“, although warning that they had probably been chosen and were not representative of the output usual. That did not stop the narrative: for weeks, the dominant conversation in the specialized media was that film studios were facing an almost perfect replacement tool: synthetic actors, sets generated in seconds, automated post-production… The Hollywood unions, which they had signed agreements with the studios the previous year after a historic strike they put the issue back on the table. Two bombs. Sora’s story has two moments of media panic, separated by eighteen months. The first arrived in February 2024, with the presentation of the model described above. There was talk that Hollywood had a serious problem, that the almost perfect replacement tool already existed and that the studios were not prepared to face this threat. The second came with the launch of Sora 2 in September 2025with real faces inserted in videos generated by AI and with third-party intellectual property by default, unless the prompts expressly requested otherwise. All of this multiplied the volume and intensity of the alarm in Hollywood and the media. What was said In February 2024, coverage of Sora’s first model mixed amazement and alarm in similar proportions. Fortune commented that OpenAI had moved the generative AI battle directly to Hollywood. NBCNews asked filmmakers if this was the end of Hollywood, and some responded that it wasn’t yet. IndieWire He sensed that Sora could mean the apocalypse of cinema. The cycle of apocalyptic headlines with Sora 2 was much more intense. CNBC declared that the app was challenging Hollywood and causing panic in the film industry. deadline He said Hollywood was raw. LA Times He spoke of a battle that was worsening and a firestorm unleashed in the sector. slatewell, he talked about how AI was about to crush Hollywood as we had known it. What happened then. The panic increased in December 2025, when Disney, the most careful entertainment company in the world with its intellectual property, signed a three-year agreement with OpenAI: investment of 1 billion dollars and access to more than 200 characters from Disney, Marvel, Pixar and ‘Star Wars’ so that Sora users could generate them in their videos. Disney+ would broadcast a curated selection of that content. It was the definitive legitimation, which has only lasted 90 days. OpenAI has closed Sora before a single dollar has changed hands. Property problems. Sora’s problems have not only been financial. The app has accumulated a long list of controversies: deepfakes of deceased public figuresmassive use of copyrighted characters without permission prior, and the appearance of external tools to remove watermarks that identified AI-generated content. In November 2025, CODA (Japanese association representing, among others, Studio Ghibli and Square Enix) sent a formal letter to OpenAI demanding that it stop using its intellectual property to train the model. The families of Robin Williams and George Carlin They publicly asked for it to be blocked generating videos with your images. Moderating generative video content at scale turned out to be much more complex than moderating text or image. The consequences of hype. Analyst Ed Zitron criticized this attitude of the media, stating that they did not cover the launch of Sora but rather they amplified their marketing. Saying that Sora was a real threat to Hollywood was, from the beginning, an extrapolation built on selected demos and clips of a few seconds. Thousands of audiovisual professionals spent months convinced that their industry was about to be replaced by a tool that, according to OpenAI’s own numbers, never found enough users willing to pay $200 a month for it. The hype cycle has real consequences: it inflates expectations that are not met, generates costly defensive decisions, and when the product closes, no one takes critical stock. Sora’s coverage is a textbook case of how uncritical amplification of tech demos can be confused with industry analysis, and the damage that attitude can do. Hollywood is still alive. The closure of Sora does not erase the generative video sector in one fell swoop: runwaywhich rejected an acquisition offer from Meta, currently leads the sector with its Gen-4.5 model, along with I see 3.1 from Google and Chinese models Kling and Seedance. These tools are absorbing the space that OpenAI abandons. Who no one absorbs is Hollywood. The film industry, with all your problems (reorganizations, box office decline, threat of streaming), remains a profitable business built on decades of well-established creation, distribution chains and franchises that no generative model can replicate with a prompt. The question is not whether AI will transform audiovisual production (it is already doing so, in post-production, pre-visualization and marketing content creation) but in what real time frames and under what viable economic models. For now, the market responds that generating photorealistic video on a massive scale is computationally very expensive and that consumer users are not willing to pay what it costs. Disney signing Sora wasn’t evidence that Hollywood was in danger. It was, rather, evidence that big studios want to be in the AI ​​conversation, not outside of it. In Xataka | Seedance’s strategy was to copy first, go viral later and back away later. Until Hollywood said “enough”

Doomsday’ are released at the same time and Hollywood already wants to sell us a new Barbenheimer. But it’s not the same

by

Warner Bros. and Disney are going to release their two big bets of the year (‘Dune: Part Three’ and ‘Avengers: Doomsday‘) on the same date: December 18, 2026. The industry has not seen something like this for years. In fact, we haven’t seen it since the Barbenheimer phenomenon, and the question is whether its effect at the box office can be repeated… when the two films share exactly the same audience. Dunesday, or whatever you want to call it. The two most powerful Hollywood studios, two of the most popular franchises in recent years, a single date. The industry already has a nickname for the event: “Dunesday“. But the fact that both coincide on that date does not respond to an agreed strategy. ‘Dune: Part Three’ arrived first in the pre-Christmas gap. ‘Avengers: Doomsday’ was initially scheduled for May 2025, but postponements derived from the strikes of actors and scriptwriters pushed it to where it is now, from where it is unlikely to move since on websites, networks and others there is a countdown in motion until the inevitable premiere. The Barbenheimer precedent. Summer 2023 was left for the annals thanks to the coincidence of two very different films, ‘Barbie’ and ‘Oppenheimer’, on the same weekend in July. The combined weekend exceeded $300 million in the US alonebecoming the fourth best in historyonly behind the three opening weekends of ‘Avengers: Endgame’, ‘Infinity War’ and ‘Star Wars: The Force Awakens’. Different audiences, even different motivations for going to the movies, but a curious, almost miraculous real possibility of doing a very rare double program on the same day. Why Dunesday is different. ‘Dune: Part Three’ and ‘Avengers: Doomsday’ don’t work like that. Both target similar viewers: devotees of science fiction and big-budget action, with a predominantly male base (this is important). Significantly or coincidentally, both feature Florence Pugh in their casts, which illustrates the extent to which they are not identical films, but do have their roots in a shared territory of blockbuster Hollywood. The idea of ​​Dunesday, as Variety says, is not to propose two different and complementary plans, but to fill the respectable with a giant dose of similar things. Why don’t they wiggle. That neither film wants to move from that date has a concrete explanation, prior to the invention of Dunesday: there are not many good gaps in that part of the year. Sony releases ‘Jumanji 3’ on December 11, a week early. Further back, the first weekend of December is usually a black hole at the box office (families are shopping, not in theaters), and Thanksgiving is traditionally occupied by Disney’s animated bet, because it is a more family-friendly box office weekend. December 18 is the best date available for two films that need a strong start. What if it works? If the experiment works, the combined box office could be among the best weekends in history. In any case, here we find the usual problem: that July 2023 generated a social phenomenon (costumes, double sessions planned weeks before, debates on networks, friendly rivalry between the actors of the films, who encouraged people to go to both) that was organized outside the industry. “Dunesday” arrives much more in advance and with a slight air of a commercial montage. That’s not going to make the films click, of course, but the truth is that we are dealing with phenomena with nuances that go in opposite directions. In Xataka | The reality of Spanish cinema: ‘Torrente’ has brought more people to theaters this weekend than any film since 2019

Lockheed has created an underwater drone that clings to ships like a lamprey. And when released, it launches torpedoes

by

The lamprey is a fish that has survived 360 million years thanks to a simple strategy: sticking to its prey to suck its blood. Lockheed Martin has taken that idea literally to name its new weapon, and the analogy is quite literal. The new thing from Lockheed is called Lamprey Multi-Mission Autonomous Undersea Vehicle (MMAUV). It is an underwater drone just over 7 meters long, capable of traveling attached to an allied ship or submarine with a lamprey-like system. While attached to the host ship, it can recharge its batteries using its built-in hydrogen generator. Stealth or attack The Lamprey MMAUV does practically everything, although it is primarily designed for covert missions. It can remain on the seabed, monitoring the enemy without being detected thanks to its acoustic signature profile. practically invisible when sonar. When the time comes to act, the Lamprey can do almost anything: it deploys decoys to confuse the opponent, it is equipped with anti-submarine torpedoes and, if it rises to the surface, it can also launch aerial drones. What makes the Lamprey especially striking is that it concentrates in a single system capabilities that until now were distributed across different platforms: surveillance, anti-submarine warfare, deception, attack and aerial reconnaissance. It can operate in a swarm coordinating with other unmanned systems. And it can do so autonomously, making decisions without direct human intervention. Autonomous submarines The Lamprey will not be the United States’ first unmanned underwater vehicle. There are antecedents like Boeing Orca submarinewith the difference that it cost eight years and 885 million dollars to develop it, all so that today it is not clear if it will end up becoming a program in the US Navy. The Lamprey has been funded internally, which Lockheed vice president Paul Lemmo said has allowed them to “iterate at lightning speed and deliver to the Navy a truly multi-purpose weapon that detects, disrupts, deceives and attacks on its own.” Furthermore, he presumes that Its cost is significantly lower than that of other manned platforms. But the United States is not the only power exploring unmanned vehicles. China has been developing its own fleet of underwater drones for some time and at the military parade in September 2025 presented the AJX002an unmanned underwater vehicle between 18 and 20 meters capable of operating autonomously, laying mines and networking with other attack systems. In Xataka | The US wants to give up bringing the most valuable samples collected on Mars. Lockheed promises to do it for less than half Image | Lockheed

In 1977 Japan released an anime inspired by a raccoon. To this day he continues to pay the consequences

by

What harm could a raccoon? Any search surface on the Internet reveals its many aesthetic virtues. They are small, but not too small; hairy, but not in moderation; intelligent, but still simple; handsome, still goofy. The dream of any child, the object of desire of every human passionate about terrestrial mammals Appearances are often treacherous. Numerous testimonies and graphic documents support the disruptive nature, in criminal occasionsof raccoons. Its own genes give it away: if its gigantic dark spots around its eyes function as a mask, the raccoon is the caco of nature, an extremely skilled animal, elusive, sagacious in its objectives, diligent in its blows. They know it well conservation services Madrid. Since the small bug was introduced into the community at the beginning of the last decade, it has spread across three different watersheds. During the last fifteen years more than 800 copiesa modest sample of a probably millennial population. They have become in a nightmare. Without natural predators (they come from the American continent), they wipe out numerous local species and cause fear among peripheral neighborhoods. The extreme expertise that only millennia of plunder provides is combined with a totalitarian reproductive capacity to dominate virgin lands in a matter of decades. The raccoon is a colonizing weapon perfect. (Thomas Despeyroux/Unsplash) We know it today, however. Half a century ago, as in many ways still today, the image of such a friendly animal conquered the hearts of a nation at the other (literal) end of the Western cultural world: Japan. A counterproductive obsession Their love-hate story begins in 1963, when American author Sterling North published Rascal: A Memoir of a Better Eraa small children’s story in which he surfs the waves of nostalgia in the company of his domestic raccoon. The work becomes an instant classic, hitting the shelves of thousands of children across the country. His media epic would enjoy a definitive boost when six years later Disney gained access to the rights to the work. Rascal, the moviewould debut in American theaters during the summer of 1969. Without viewing, the film would contemporize the dazzling success of the friendly raccoon in the United States, and limit its legacy. Until 1977. Almost fifteen years after its publication, Nippon Animationa Japanese animation studio, had an idea: how about moving the story of Rascal to the small screen, in a production of 52 episodes intended for family consumption? Overnight, Rascal, its irresistible manga version, conquers hyperbolic Japanese pop culture. It is difficult to define the impact of the series. Rascal would end up appearing in television advertisements and video games intended a la GameBoyand would cause thousands of Japanese children to want a raccoon in their homes. What harm could the proverbial Rascal do, after all? It was 1977 and Japanese parents had no choice but to shrug their shoulders. In the blink of an eye Japan started to matter raccoons like there was no tomorrow. The fever reached its peak in the late seventies, when Japanese families acquired the mammalian sibylline at a rate of 1,500 copies for weeks. Suddenly, Japan had placed a Trojan horse perfect in its natural ecosystems. And he had done it driven by an animated series. And the raccoons took over Japan The consequences were quickly felt. How do they explain in Atlas Obscuraone of Rascal’s moral readings was the liberation of the animal. Raccoons, after all, are wild animals, and at the end of the day they only want one thing: to flee. The idea fit well into the Japanese cultural world, soon to any symbiosis spiritual between fauna and flora. Many Japanese parents learned the lesson the hard way: the raccoons had begun to behave like, err, raccoons. Aggressive, destructive and difficult to domesticate, many of them were found where the fable of Rascal entrusted them: in nature. Turned into a nightmare, the series offered a comfortable moral safeguard. The subsequent history is similar to that of Madrid. Within a handful of years raccoons had spread throughout Japan. At the end of the last decade, its presence was known in no less from 42 prefectures (out of a total of 47). They looted templesthey finished with species natives with similar characteristics (the tanuki) and disrupted numerous ecosystems and crops, generating annual damages worth €300,000. The Japanese government would not take long to prohibit the importation of raccoons, imposing severe fines on anyone who dared to go to the black market, but the damage would already be irreparable. The raccoon continues to roam freely in the archipelago, and Rascalvery oblivious to the consequences caused by his media enthronement, remains very popular. The beginning of the end. Even though the raccoon has sneaked in in many nations of the planet (Germany catches about 25,000 every year), only in Japan does its history rotate around pop mythomanias and animated series. Its presence is probably irreversible. As this report As Slate illustrates, the raccoon is not only an animal suitable for the countryside: it is also a nearly perfect urban pest. His grasping hands allow him to avoid countless traps, and his particular intelligence causes the policies to stop him to become obsolete in a matter of days. Cities, in essence, function as a field of military training. Each obstacle posed by public authorities offers valuable learning that always ends up being overcome, and that underpins the adaptability urban of the species. In Toronto, for example, the introduction of famous anti-raccoon garbage containers, supposedly impassable, was revealed useless after two years. Nothing that the Japanese governments don’t know about. Thank you, Rascal. Image | Richard Burlton In Xataka | We have found an ancient bone in Córdoba. Some believe it is part of Hannibal’s war elephants. In Xataka | 13% of Spaniards have tried cocaine once in their lives. If we ask the dogs of Madrid the percentage will be higher

Filmin has released a documentary about the riot police of the process. And now it has threatening graffiti on its headquarters

by

The Barcelona headquarters of Filmin It woke up on January 20 with graffiti on its façade: “Collaborators with Spanish repression.” The message, signed by the independence collective Nosaltres Sols!, marks the most critical moment of a boycott campaign that began days before on social networks. The trigger: the programming of the documentary ‘Icarus: the week in flames’, focused on the testimonies of riot police from the National Police who acted in Barcelona during the conflicts of October 2019, after the sentencing against the leaders of the processes. What is it about? ‘Ícarus: the week in flames’, directed by Elena G. Cedillo and Susana Alonso, reconstructs the riots that occurred in Barcelona for seven days in October 2019, after the sentence against the leaders of the processes. The documentary, filmed in 2022 and available on Filmin since January 9, is based on interviews with agents and commanders of the Police Intervention Units who participated in the operations. “We had the feeling that this was a war,” declares one of them. The sequences include material recorded by the riot police themselves and from a helicopter, with scenes of the clashes in El Prat, Urquinaona Square and in front of institutional headquarters. The answer. Jaume Ripoll, editorial director and co-founder of Filmin, has tried to defuse the controversy appealing to the classic principle of “Programming a film is not equivalent to subscribing to its approach.” The platform insists that it does not censor content based on ideological orientation and defends that cinema should serve to “look squarely at what makes us uncomfortable.” However, the virulence of the reaction raises a question that transcends the specific case: can streaming platforms maintain a position of editorial neutrality, or does their catalog inevitably reflect an ideological position? Filmin is the platform with greatest historical commitment with the Catalan language and culture. In June 2017, two years before the events narrated ‘Icarus’the company launched Filmin.catbecoming the first digital platform for series and movies specifically in Catalan, anticipating giants like Netflix or Disney+. According to the last report from the Audiovisual Consell of Catalonia Introduced in December 2025, Filmin includes Catalan (in audio, subtitles or both options) in 2,350 titles in its catalogue, which represents 20.7%. The figure contrasts radically with Prime Video (9.5%), Netflix (3.5%), Max (3.2%) or Disney+ (2.2%). The paradox. The data makes the controversy more striking: Filmin is the platform that has historically supported the Catalan language and culture the most. In June 2017, two years before the riots that ‘Ícaro’ documents, the company launched Filmin.catthe first digital platform dedicated specifically to series and cinema in Catalan, before Netflix or Disney+ did so. Latest report from the Audiovisual Consell of Cataloniafrom December 2025, places Filmin as the platform with the greatest presence of Catalan (whether in audio, subtitles or both) with 2,350 titles, 20.7% of its offer. The figures for Prime Video (9.5%), Netflix (3.5%), Max (3.2%) or Disney+ (2.2%) are much lower. The platform has also produced original fiction in Catalan through Filmin Originals, such as ‘Selftape’, a series by the Vilapuig sisters about abuses in the Catalan audiovisual industry, and has co-produced titles with international coverage such as ‘Molt Lluny’ or ‘Forastera’. This history makes more surprising the painting of “collaborationism with Spanish repression” that it has received, signed by Nosaltres Sols!, a right-wing independence group born from the 2019 mobilizations led by the influencer David Silvestre. Other platforms and ideological controversies. The case of Filmin is not an exception in the sector. The large platforms have experienced comparable situations in recent years: debates about the permanence of certain content in the catalog and whether programming a work implies endorsing it. Netflix went through one of its greatest internal turbulences in 2021 due to the comedy special ‘The Closer’, by Dave Chappelle, which sparked accusations of transphobia. Trans workers from the company called a protest rally calling for the withdrawal of the program. Ted Sarandos, co-CEO of Netflix, justified keeping it on the platform by arguing that “not all of Netflix’s content will be to everyone’s taste.” Woody Allen, of course. Prime Video faced a conflict of a different nature in 2019, when Woody Allen took the platform to court for breaking a contract that provided for the distribution of four feature films. Amazon argued that the “public perception” of the filmmaker had changed after accusations of sexual abuse against him reappeared, a circumstance that made the agreement unviable from a commercial point of view. The dispute ended with an extrajudicial agreement whose terms were not disclosed, but established jurisprudence: platforms can disassociate themselves from commitments with creators if they consider that their reputation damages the corporate image, without the need for judicial convictions. And ‘Gone with the Wind’. HBO Max adopted a measure in June 2020 that caused international reactions: the Temporary withdrawal of ‘Gone with the Wind’ coinciding with the moment of greatest intensity of the Black Lives Matter mobilizations. The platform explained that the 1939 film reproduced “ethnic and racial prejudices” that could be “hurtful” seen from the present. Weeks later, the film returned to the catalog preceded by an intervention by film historian Jacqueline Stewart, who contextualized its historical relevance while pointing out its racist representations. The formula chosen by WarnerMedia established a middle path: a title can remain available without omitting its conflicting elements, as long as they are presented with the necessary critical framework. Header | Jaume Ripoll In Xataka | Disney+ has discovered that Generation Z does not want to watch its two-hour movies. So he’s going to give them vertical microdramas

In 2021, BBC released a video about China causing an earthquake. Now it’s a meme that glorifies Chinese cities

by

Trends on social networks are, in many cases, inexplicable. Overnight something goes viral and it’s easy for us to not even know where it came from. In the summer of 2025, chinese networks began what from the West we could see as a simple memeeven nonsense: many videos that show panoramic views of Chinese cities to the rhythm of the mythical BBC intro. This meme spread and is useful for observing some of the most impressive cities in the world from a drone view. There are even users commenting on how some cities, like Chongqing, had undergone a radical transformation in just 20 years. The videos, without a doubt, are impressive and there is a example after other…and after other. But behind the meme there is something much more interesting: an outbreak of international conflict because of… the BBC. BBC News countdown intro style meme continues in China. Below in order is for Guiyang, Nanjing, Jinhua and Jieyang. https://t.co/EKZopt48Pc pic.twitter.com/LhjHVATMKW — JR Urbane Network (@JRUrbaneNetwork) September 1, 2025 The BBC video that angered 1 billion people In February 2021, the world was still reeling from the aftermath of COVID-19. Wuhan, the Chinese city identified as the focus of the global pandemic, was a monitoring point for world news due to the government’s policies to fight the virus. And the BBC published its controversial ‘How everyday life has changed in Wuhan’. It’s this video: Up to this point, we might think that it is just another report, but they published it in duplicate. The one above is the international version, in English. The one I leave you below is the version for China: Have you noticed any difference? Let’s go with some screenshots: International version Chinese version International version Chinese version International version Chinese version International version Chinese version Already we saw it in Xataka back in the day: The international version has a gray filter, while the Chinese version shows more vivid colors. That, without us realizing it, creates a narrative. And those who did notice were some Chinese Internet users and the state media Global Times. Chinese social networks named the filter used in the international version as “underworld filter” or “gloom filter”but the one who gave it the most importance was the aforementioned state tabloid. He accused the BBC of adding greyish filters to its reporting on China to make the country appear dystopian and polluted. It did not stop there: the matter spread like wildfire on networks and the tension escalated to the point that the international broadcast of BBC World News was banned in China that same month. In fact, international spokespersons have on occasion used the hashtag #GloomFilter to criticize Western coverage of China. The BBC defended its editorial independence, rejecting accusations of bias, but both the BBC and Chinese media have since starred cross attacks. A lot has rained since 2021 and, as I pointed out at the beginning of the article, it is now meme stuff. The BBC intro accompanies luminous images of Chinese cities without the “underworld filter.” And it is an example of how something that, at first glance, may be a story without much history, hides much more. And, well, the story of Global Times throwing darts at the BBC did not end in 2021, but has lasted until recently, mentioning that “BBC has become one of the most destructive negative examples in the global media landscape.” But beyond all this, the truth is that the videos are impressive, showing dystopian cities in some cases. Images | BBC In Xataka | China loves Europe so much that it has built its own: these are the replica cities that populate the country

AEMET has released its prediction for winter and confirms the trend that is no longer an anomaly: a winter “without cold”

by

Although we can keep in mind that winter does not begin until next December 21, coinciding with the winter solsticefor meteorology now we have started with the station from today. A season in which we could all expect a great spell of polar cold to be at home with a blanket and watching a series on television. But the AEMET has lowered these forecasts taking into account to what we experienced in previous years. Via a post on X The AEMET has welcomed this new winter 2025-2026, but with bad news behind it: it will be much warmer than usual with a high probability. We are not talking about individual “summer” days, but rather a robust statistical signal that covers the entire quarter (December-January-February). What we used to call an anomaly, the data are beginning to call the norm: winter in Spain is fading. Heat map. AEMET’s seasonal prediction It doesn’t leave much room for doubt. According to probabilistic models, the average temperature will be in the warm zone throughout the country. Specifically, for the AEMET the eastern peninsula and the Balearic Islands have a probability of a much warmer winter that exceeds 70%. In the case of the Canary Islands and the Balearic Islands, this is where the “zero zone” of this warming will be found, with a very pronounced thermal increase with respect to its normal values. In the rest of the peninsula, the probability is around 50%, which continues to be a sign that points to having a winter that is as normal as possible with respect to what we have seen in previous years. The rain. If in terms of temperatures it seems that we are not going to have very good news with a high probability, in terms of precipitation it seems that we must be optimistic. A priori, the models suggest that we will not have an extremely dry winter but nor will it be too wet. And the rainfall seems to be close to the average, although with great variability. Not all months of this winter will rain in the same way, emphasizing especially the second half of winter, that is, the end of January and February, where the models point to the arrival of dynamic phases with fronts and storms. This is something that may fit with studies on the loss of sea ice in the Arctic, which alters atmospheric circulation and may lead to much more “wet or variable” winters in the Iberian Peninsula, breaking the patterns that we saw in our environment. 28 days of “no winter”. To understand why the AEMET is so sure of this forecast, you have to look in the rearview mirror. The most recent reportslike Climate Central, already warned that last winter Spain experienced an average of 28 days with temperatures above the historical average. To do this, experts focus on reducing the days where we have temperatures below zero with a sharp drop in the days where there is frost. Furthermore, cities like Valencia are seeing how urban centers are turning into ovens even in the middle of winter. And it is a serious danger, as the CLIVAR-Spain report warns that this amplification of warming and the alteration of winter variability pose a critical challenge for our ecosystems, which need rest from the winter cold for their biological cycles. Goodbye to the historic cold. What AEMET is telling us with this forecast for 2025-2026 is that the atmosphere in Spain has more and more accumulated energy. Studies by Funcas and analysis by AEMET itself corroborate that the decrease in snow coverage and the increase in warm episodes are not temporary, but in the end they are the reality we face. We are facing a scenario where winters do not disappear, but they do “soften” until they become unrecognizable compared to those of three decades ago. If you have thermal clothing prepared for this year, it is possible that, except for occasional episodes of storms in February, it will stay in the closet. Images | Thomas Holmes Immo Wegmann In Xataka | “Three days of pure cold”: while the world looks at the polar vortex, bad news accumulates for AEMET

OpenAI has released GPT-5.1 with two personalities because 800 million users do not want the same AI

by

OpenAI has launched GPT-5.1an update of its flagship model that comes in two variants: Instant (conversational and “warmer”), Thinking (deep reasoning). The real novelty is not in the technical metrics, but in something more prosaic: you can now choose between eight conversation tones, from “professional” to “cynical.” It is recognizing that AI as a mass product needs segmentation. It is no longer enough to have just one assistant for everyone. The new model selector for Plus users. Image: Xataka. Why is it important. OpenAI has 800 million users with radically different expectations. Some want a neutral and efficient assistant. Others seek warmth and empathy. Some have even developed problematic emotional ties to the chatbot. The company tries to solve this with personality adjustments, but the underlying problem remains: ChatGPT keeps pretending to be a person, a consistent entity that knows you. This generates the same risks of emotional dependence that have motivated mental health demands and alerts. The facts: GPT-5.1 Instant improves in math and programming, and for the first time uses “adaptive reasoning”: decide when to think harder before answering. GPT-5.1 Thinking, for its part, dynamically adjusts its processing time according to the complexity of the question, being twice as fast in simple tasks and twice as slow in complex ones. The eight available tones (Default, Professional, Friendly, Sincere, Quirky, Efficient, Geek, Cynical) work by injecting different instructions into each prompt. The capabilities of the model do not change, only the presentation changes. Yes, but. The speed of the launch has come at a cost. OpenAI itself admits in its technical documentation that GPT-5.1 presents “known security regressions” compared to the October version. They prioritized time-to-market over exhaustive testing, something striking in a company under intense regulatory scrutiny due to cases of vulnerable users. Furthermore, personalization has limits that OpenAI has had to explicitly acknowledge: “taken to the extreme, personalization would be useless if it only reinforced your worldview.” It is admitting that you are walking a tightrope between engagement commercial and social responsibility. Between the lines. The launch of GPT-5.1 is a symptom of a deeper strategic shift. OpenAI is fragmenting its product because the “one AI fits all” model has failed. GPT-5 was so disappointing that the company had to enable it again GPT-4o as an option the next day. In Xataka | OpenAI has never been more ambitious. And he’s never been so close to not being able to pay his debts. Featured image | Xataka with Mockuuups Studio

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.