Model

We believed that no Chinese AI model would soon come close to Fable 5 or GPT-5.5. Then GLM-5.2 arrived

by

A few days ago, the Chinese startup Zhipu AI (Z.ai) announced the launch of its new open AI model, GLM-5.2. It did so boasting amazing features that brought it very close to the best closed models from OpenAI and Anthropic, something that seemed impossible. Well, the more analysis is carried out on the model, the better off it is. We may be at the beginning of something very important. A change of trend. GLM 5.2. The Chinese startup Z.ai has been releasing different versions of its GLM AI model for a long time, but the latest one is undoubtedly the most surprising because its performance is especially promising. It has 744,000 million parameters (744B), of which 40,000 are those that remain active. We are looking at a model with a context window of one million tokens and a new architecture called IndexShare/IndexCache. Better than GPT-5.5, very close to Opus 4.8. The startup showed how the performance of GLM-5.2 is extraordinary in programming tasks. In the FrontierSWE test, the most demanding of those currently available, GLM-5.2 outperformed GPT-5.5 and only Opus 4.8 was superior by a very small margin. The same happened with other tests such as PostTrainBench or SWE-Marathon, which, for example, evaluates the behavior of the model in very long autonomous programming sessions. Source: Z.ai. In many other tests the photo was identical: the model has made a spectacular leap since version 5.1, and is in many tests almost as good (or better) than the best from OpenAI, Anthropic or Google. But it’s not just them who say it.. Artificial Analysis, a reputable independent firm that maintains an updated ranking of the performance of the new AI models that are arriving on the market, confirms the data of Z.ai itself. In his tests he indicates how the “intelligence index” of GLM-5.2 is now 51 points. It is only surpassed by GPT-5.5 (55), Claude Opus 4.8 (56) and Claude Fable 5 (60). Source: Artificial Analysis. This Chinese open model leaves behind the new Gemini 3.5 Flash, but also Chinese competitors such as Qwen 3.7 Max, MiniMax-M3 or DeepSeek V4, among others. The jump in quality from GLM-5.1 is, we insist, outstanding, much greater than what, at least according to this index, was seen from Opus 4.8 to Fable 5. The jump in performance is spectacular, although it is true that the comparative price to solve the tasks proposed in the benchmark rises significantly. Source: Artificial Analysis. But it’s not perfect. The Artificial Analysis report, however, shows that although GLM-5.2 is very strong in areas such as programming, it is weak in others. For example, it is far from being as reliable as Fable 5, GPT-5.5, Claude 4.8 or Gemini 3.1 Pro in terms of correct answers, which is also lower in proportion to that of its competitors. However, his hallucinations have significantly reduced. And it’s much (much) cheaper. But in addition to being fantastic in many areas, it is much cheaper than its competitors. Maintains the price per million input/output tokens of its predecessor ($1.4/4.4), while that of GPT-5.5 It’s 5/30 dollars and that of Opus 4.8 10/50 dollars. It is true that it consumes many more tokens than GPT-5.5 (very efficient) or Claude Opus 4.8, but even with that its final cost is much lower. My tests with GLM-5.2 programming. I’ve been a Z.ai subscriber for months now because they offered an annual subscription at the end of 2025 at a really low price. This has allowed me to test GLM-5.2 for a few hours and although I cannot draw definitive conclusions, it does seem clear that there is a leap in quality in terms of its ability to program. I asked him to review a personal code project and he identified several security flaws and possible improvements in great detail. Chatting with GLM5-2. In conversational mode the behavior is much more difficult to evaluate: I have been interacting with the model and asking it questions, and although it is better than GLM 5.1 many times, other times it is not so much and I would say that in terms of creativity to write the frontier models of Google, OpenAI and especially Anthropic they are still quite superior. You can try it on their websiteand there you will see something else: it takes significantly longer to respond than other chatbots, because its reasoning phase is longer. Take more time to answer questions. Benchmarks are one thing, experience is another.. In the absence of testing it (much) more, of course the impression is that the model has improved significantly compared to a GLM-5.1 that had lagged behind its Chinese competitors (not to mention the current Claude Opus 4.8 or GPT-5.5). On platforms like Reddit opinions are dividedbut many consider it a fantastic option to run locally… if you have a very, very powerful machine with at least 256 GB of unified memory (Mac Studio). And one thing seems clear: when using it as an AI model for programming, comes surprisingly close to Claude Opus 4.8. In Xataka | Chinese technology companies entered the AI ​​race with cheaper models than the rest. That’s starting to end

everything we think we know about Apple’s new base model

by

In September the new iPhone 18 Pro and it will be the debut of John Ternus as the new CEO of Apple. And… that’s it. Maybe we’ll see the rumored foldable iPhonebut the iPhone 18 will have to wait. Breaking with the tradition of their releases, the basic iPhone 18 will have to wait until sometime in the spring of next year. However, that does not mean that the wheel of rumors and leaks is not working. Next, we tell you everything we think we know about that iPhone 18 which, according to those rumors, will not be a revolution, but it will have a couple of features that will make it more interesting than the current ones. iPhone 17 and iPhone 17e: new camera and RAM memory. Let’s go to trouble. What to expect from the iPhone 18? Design: two aspects in the leaks with a design maintained in the most conservative leak and an island that will follow the design of the Pro models, but with two cameras instead of three to further differentiate this iPhone 18 from the iPhone 17 and iPhone 16. Screen: 6.3-inch diagonal with LTPO refresh rate up to 120 Hz with a Dynamic Island that will maintain the size of the one we had until now. On the iPhone 18 Pro we are supposed to have a smaller Dynamic Island. SoC: Apple A20 with reduced features in GPU and cores compared to the A20 Pro that the iPhone 18 Pro will carry. RAM: 12 GB integrated on the chip wafer to allow good performance of the new Siri AI and Apple Intelligence. Cameras: dual camera configuration with a wide angle that will be brighter and so on, but that will pale in comparison to a new main sensor that will also be in a camera with a variable aperture. Connectivity: The new C2 chip would be the heart of the networks of the new iPhone 18. Price: It is expected to remain around what we already have with the iPhone 17, without exceeding 1,000 euros in the basic 256 GB version. Another story will be the version with 512 GB, which may suffer due to the increase in the price of components. Launch: Obviously, without anything confirmed, but all the leaks for a long time point to a launch in spring 2027. When would the iPhone 18 come out? The forecast is that Apple, for the first time, will break its classic release cycle. Instead of launching the two iPhone 18s in September, the one that will arrive first will be the iPhone 18 Pro, while the iPhone 18 will be released sometime in spring next year. In fact, this late launch would imply that in the September keynote Apple will not present the iPhone 18 in depth. It should mention it so that the user is clear that it will arrive, but leaving all the details for a later presentation closer to the launch in spring. What design will the iPhone 18 have? The leaks point to a conservative design, as usually happens. Apple has already created a visual identity for the Pro with a giant side-to-side rear camera island, while the basic iPhone 18 has the cameras separately and vertically. There may be some changes such as cameras next to each other or with a similar design to the Pro’s island, but keeping only two cameras. At the moment, there is only speculation here, but if the trend continues, it is easy to see that change on the back because, if not, Apple would repeat the design of the iPhone 16 for the third year. And that is not something that Apple people usually do. What will the screens of the new iPhone 18 be like? On the front, the iPhone 18 Pro is expected to feature a smaller dynamic island, but right now there are no leaks about that element of the iPhone 18. There have been reported some problems when carrying out this miniaturization, which would raise the price of the device and is something that, for the ‘cheaper’ iPhone 18, Apple would not want to allow itself. For the rest, the leaks point to an LTPO screen with a refresh rate up to 120 Hz and a diagonal of 6.3 inches. There are fewer leaked details (in general) than for the iPhone 18 Pro, something logical due to this supposed time lapse between one model or another, and it’s not like we have too much information on the screen either. The easiest? That the Pros do have that smaller Dynamic Island, but that with the iPhone 18 Apple maintains the size. What processor will the iPhone 18 have? Here what we can expect is a A20but with some cores and frequencies cut compared to the A20 Pro that, supposedly, will mount the iPhone 18 Pro. It will be a 2 nanometer chip manufactured by TSMC and, beyond the expected increase in power thanks to the new SoC, perhaps the most notable thing is the issue of RAM. From 8 GB, we would go to 12 GB of memory. They are the same ones that already have the iPhone 17 Prohe iPhone Air and those expected to mount the iPhone 18 Pro and the explanation when it comes to matching the amount of RAM is marked by the company’s ambition with the new Siri AI. The iPhone 18 will be mobile phones launched with the new Siri and Apple Intelligence in place and, to perform some actions locally, a considerable amount of RAM is necessary. It is clear that 12 GB is needed to have all the Siri AI and Apple Intelligence options The iPhone 17’s 8GB is adequate for certain tasks, but not for the more advanced Siri AI tasks (as Apple itself has detailed) and going up to 12 GB would be ideal to maintain parity in that experience with that AI that they are going to push so much from now on. What battery will … Read more

list of new features of the new version of Anthropic’s Artificial Intelligence model

by

Let’s tell you What’s new about Claude Fable 5, Anthropic’s new public model for your chat artificial intelligenceThis is the first model of the new Mythos classwhich already when it launched its preview version two months ago did so saying that it was so powerful that they would limit its power. Anthropic has launched two new models, the Claude Mythos 5 which will be available only to “a small group of cyber defenders and infrastructure providers”, and a Claude Fable 5 more adapted for mainstream users. It is a model with all the capabilities of Mythos Preview, but with security measures so that it is not used for bad things. And precisely because it is the model that is reaching users, it will be precisely Claude Fable 5 that we are going to focus on. we will give you a list with everything that changes so you know what it’s like to use compared to the previous version. Claude Fable 5 news Mythos-level capabilities for everyone: Fable 5, along with Mythos 5, is the most powerful AI model in history. Their capabilities are enormous, with performance that makes them lead the test benches. It is the first that allows you to use the capabilities of Mythos 5, and the jump from Claude Opus 4.8 is truly surprising, leaving GPT-5.5 and Gemini 3.1 Pro far behind. Longer freelance work: Fable 5 can work autonomously for longer than their predecessors. This makes it possible to tackle longer and more complex tasks that could not be sustained before. A big leap in programming: Programming is another aspect where this model has improved the most. Anthropic highlights the case of Stripewhich during initial testing performed a complete migration of a 50 million-line Ruby code base in one day, a job that would have taken a human team more than two months. More efficiency: Fable 5 is also more efficient in token consumption than previous Claude models, good news for teams who want to get the most out of it. Improvements in knowledge work: Fable 5 shows solid performance on complex analytical tasks. Anthropic claims to outperform the competition in senior-level reasoning, with notable improvements in reasoning about documents, interpreting graphs and tables, and problem solving. Great improvement in vision tasks: This new model can extract concrete figures from detailed scientific figures and reconstruct the source code of a website based solely on screenshots. For example, he was even able to beat Pokémon FireRed based on vision alone, while previous models had trouble playing even with assistive tools. Better memory and long context: The model is able to not get lost over millions of tokens, maintaining focus on the context, and improving memory. Something perfect for longer and more complex tasks. Maintains its barriers against misuse: Anthropic also says its new model keeps the level of misaligned behavior. This means that deceiving him or trying to get him to cooperate in misuse is still just as difficult. Safeguards with forwarding to Opus 4.8: Fable 5 is based on Mythos 5, which Anthropic said was so powerful it was scary. That is why it comes quite well-equipped and with many security measures. For example, if it detects that we are asking something “dangerous”, it avoids the question and even forces the use of an inferior model, Claude Opus 4.8. New data retention policy: For Mythos-class models, including Fable 5, Anthropic requires 30-day data retention for security monitoring purposes, both on its own and third-party surfaces. Of course, they say that they will not use that data to train new models or for anything that is not related to security. After 30 days, it will delete all data from our conversations in almost all cases. Price and availability: Fable 5 costs $10 per million tokens in and $50 per million tokens out, less than half of what Claude Mythos Preview cost. Developers can use it in the Claude API with the identifier claude-fable-5. Plus, it’s included at no extra cost until June 22 for Pro, Max, Team, and Enterprise subscribers. Of course, starting June 23, it will require usage credits until the capabilities of the company’s servers allow it to be incorporated as standard in payment plans in the future. In Xataka Basics | How to prevent AI from always being right by default and thus make Claude, Gemini and ChatGPT have fewer hallucinations

Claude Fable 5 is the most powerful public AI model in history. Also the most expensive, exclusive and frustrating

by

When Anthropic presented Claude Mythos Preview two months ago, he did it with a singular message: it is so powerful that you will not be able to use it. That, of course, caused everyone to want access to it. Well: Anthropic has just introduce Claude Fable 5 and Claude Mythos 5its new AI models directly derived from that. There is good news, but also bad news. Like Mythos, but capped as a precaution. Anthropic already warned that Claude Mythos Preview was a spectacular tool for finding security vulnerabilities. That made it especially juicy for cybercriminals, so the company decided that only a few trusted entities (under its Project Glasswing) would have access to the model. That learning has now been applied, because in this announcement we have two different (and layered) versions of the model: Claude Fable 5: a model with all the capabilities of Mythos Preview, but with notable security measures that prevent it from being used for malicious purposes. As soon as the model detects that we are asking something “dangerous”, it avoids the question and even forces the use of an inferior model, Claude Opus 4.8. Clear examples: questions about cybersecurity or the development of biological weapons, for example. Claude Mythos 5: This version is somewhat less capable than Fable 5 in terms of cybersecurity, but will only be available to “a small group of cyber defenders and infrastructure providers.” It is the natural heir to Mythos Preview, and according to its creators it is even better than the original version. Claude Fable 5 / Mythos 5 simply sweeps the most demanding benchmarks on the planet. There have never been more powerful models. Anthropic’s internal testing shows that we are facing the most powerful AI models in history. In all benchmarks – including the new FrontierCode programming, much more demanding than SWE Bench Pro – the scores of Claude Fable 5 and Mythos 5 are simply spectacular, well above those of their rivals. The jump from Claude Opus 4.8 is really surprising, but it leaves GPT-5.5 and Gemini 3.1 Pro far behind (they don’t compare with the recent 3.5 Flash). This is a brutal blow to Anthropic’s table, and we will see how both OpenAI and Google respond. Claude Fable 5 is amazing. Ethan Mollick, well-known AI popularizer, has had access to Fable 5 for a few days and is amazed by the experience. With this model he has managed to complete projects such as east of the isochronic map that previous models had never solved, and it has done it almost “the first time”. In one of the cases Fable 5 worked for 9 and a half hours straight to produce a code called Concord of data analysis. Their conclusions are compelling: Last year (when working with GPT-5 Pro) I called him “work with a magician”: you recite the spell and something happens. With Fable, the spell has become so powerful that I’m no longer sure I’m the wizard. I feel more like a patron. I describe what I want, pay for it and evaluate the result. The conspiracy takes place somewhere I can’t see, in hundreds of small decisions over which I never have a say. Work has gone from being a process to being a result. I no longer direct; charge. The criticism is unanimous. Andrej Karpathy, who recently signed by Anthropic, commented on X how this is a qualitative leap that for him is of the same relevance as the one that Claude 4.5 represented in November. That model began the overtaking of OpenAI: this puts it even further away (at least, for now). Other tweetersemployees or not from Anthropic, make it clear that this is an important leap in the capabilities of AI models. It’s only been a few hours since the launch, but everything points because we are indeed facing a notable leap in quality. Consume tokens like there’s no tomorrow. But in the face of that fascination, the criticism. Discussions on Reddit reveal how users who have started using it have quickly detected the problems associated with this release. The first of them: Claude Fable 5 burns tokens like there is no tomorrow. Its consumption is enormous, and the quotas for Pro and even Max accounts run out in minutes if we use the model intensively. If it already seemed to us that we were exhausting the limits of the free or quick payment accounts, with Claude Fable 5 that feeling worsens: Fable 5 is fantastic, but we can barely use it often with the Pro or Max plans because those dreaded messages about waiting X hours to continue using it quickly appear. Extremely cautious. Anthropic has been very serious about avoiding misuse of Fable 5, and as soon as it detects anything suspicious it “brakes” and “downgrades” the model so that at that moment the one that is activated is Claude Opus 4.8 (which is not bad at all). The problem is that users are detecting that the model takes completely harmless prompts as dangerous. Although in Anthropic indicate Although these security measures are activated in less than 5% of sessions, what users are detecting is that they are activated much more. Fable 5 can get silly. Not only that: Fable 5’s own design means that if it encounters a prompt that it detects as dangerous, the model tries to avoid the response and automatically reduces your capabilities (‘nerfing’) without you knowing. It gets a little sillier on purpose, so to speak. As Anthropic itself explains on the system card, We have implemented new measures that limit Claude’s effectiveness in requests related to the development of cutting-edge large-scale language models (LLMs) (for example, in creating pre-training pipelines, distributed training infrastructure, or designing machine learning accelerators). Using Claude to develop competing models already violates our Terms of Service, but enforcing this restriction through our security measures prevents giving an advantage to those users most willing to violate those terms. Unlike our cybersecurity, biology and chemistry interventions, and distillation attempts, … Read more

list of new features of the new version of Anthropic’s Artificial Intelligence model

by

Let’s tell you What’s new in Claude 4.8 Opus, the new version of Anthropic’s most advanced and powerful artificial intelligence model. This version has surprised us by arriving just 41 days after Claude Opus 4.7, and it seems that the improvements are minimal, but there is a really important change in its honesty when it comes to telling you if it doesn’t know something. In any case, here you have a complete list with all the new features that come with this new version of Claude 4.8 Opus. We are going to explain each of them briefly so that they are easy to understand. Another thing you should know is that Opus is the most advanced line of Claude models, the one indicated for more complex tasks for programming and the one that uses up your limits the fastest when you use it. There is also the most efficient Sonnet model for day-to-day tasks, which continues in version 4.6 since February 2026, and a Haiku for quick and simple questions that continues in version 4.5 since October 2025. News from Claude 4.8 Opus A more honest AI: The prominence of this new version goes to honesty. He’s significantly more honest about his own work, telling you when he’s unsure about something. It’s also about four times less likely to let bugs in code slip by without flagging them, compared to its predecessor. Performance improvements: The Agentic code score for creating code with agents increases from 64.3% to 69.2%, and multidisciplinary reasoning with tools increases from 54.7% to 57.9%. On other test benchesin the SWE-bench Verified it goes from 87.6% (Opus 4.7) to 88.6%, and in Terminal-Bench 2.1 it rises from 66.1% to 74.6%. GPT-5.5 still falls short in terminal/CLI workflows, although there have been big improvements in Claude, and both models are practically on par in web browsing and graduate-level science topics. Alignment improvements: Alignment assessments show new highs in prosocial traits such as supporting user autonomy and acting in their best interest. Rates of misaligned behavior such as cheating are lower than in Opus 4.7. Fewer hallucinations: As usual, the number of hallucinations is also reduced. Honesty when telling yourself when you don’t know something also helps reduce them. Quick mode: According to AnthropicOpus 4.8’s fast mode is now about 2.5 times faster. The company claims that the improved Quick Mode also costs three times less than before. Effort control– Users can choose between “extra” or “max” levels so that the model spends more tokens and obtains better results. Dynamic Workflows (preview for research): With this new feature, Claude can schedule work and run hundreds of subagents in parallel in a single session, being able to complete codebase-scale migrations of hundreds of thousands of lines. Available for Claude Code on Enterprise, Team and Max plans. No change in base price: The base price of API tokens is unchanged from Opus 4.7. It is 5 dollars per million input tokens, and 25 dollars per million output, with up to 90% savings with prompt caching and 50% with batch processing. In Xataka Basics | How to prevent AI from always being right by default and thus make Claude, Gemini and ChatGPT have fewer hallucinations

a 98″ model at a knockdown price

by

Xiaomi continues without giving truce in the market televisions with an affordable price. The Chinese firm, which has been sneaking into living rooms around the world for years by offering televisions with an unbeatable quality-price ratio. Now it renews its QD-Mini LED range with the Xiaomi TV S Mini LED 2026a family that for the first time in this price segment, extends from 55 inches to an impressive 98-inch model. There are five screen sizes that do not waver in their technological commitment and do so at prices that, once again, invite us to ask the usual question: how do they do it? The answer, at least in part, lies in the technology inside them. All models mount QD-Mini LED panelsthe combination of Mini LED backlight with quantum dots that allow us to get closer to the chromatic purity of OLED without giving up a brightness that organic panels cannot yet match. The result is a panel that, according to Xiaomi, is capable of reaching 1,200 nits of peak brightness, reproducing 94% of the DCI-P3 color space and reaching 0.0001 nits of pure black. Something unthinkable in this price category just a couple of years ago. Xiaomi TV S Mini LED 2026 from 55″ to 75″ Xiaomi TV S Mini LED 2026 85″ and 98″ screen QD-Mini LED 4K (3,840 x 2,160 pixels) 60/120Hz 94% DCI-P3 (typ) 308, 384 and 512 dimming zones Peak brightness: 1,200 nits 178° (H)/178° (V) MEMC: 4K 60Hz HDR10+, HLG, Filmmaker QD-Mini LED 4K (3,840 x 2,160 pixels) 144/288Hz 94% DCI-P3 (typ) 532 dimming zones Peak brightness: 1,700 nits 178° (H)/178° (V) MEMC: 4K 120Hz Dolby Vision, HDR10+, HLG, Filmmaker Dimensions and weight with base 1,667 × 391 × 1,026 mm and 23 kilos 1,445 × 391 × 901 mm and 16.3 kilos 1,226 × 312 × 769 mm and 11.7 kilos 2179 × 84 × 1,246 mm and 50.8 kilos 1,890 × 413 × 1,154 mm and 33.4 kilos CPU Quad cortex A73 Quad cortex A73 GPU Mali-G52 (2EE) MC1 Mali-G52 (2EE) MC1 RAM MEMORY 2GB 3GB STORAGE 32GB 32GB wireless connectivity Wi-Fi 5 Dual band 2.4/5 GHz Bluetooth 5.0 Wi-Fi 6 Dual band 2.4/5 GHz Bluetooth 5.2 ports DVB-T2/C, DVB-S2 1x HDMI 2.1 (CEC ALLM VRR), eARC (HDMI 2) 2×HDMI 2.0 1x USB 2.0 Ethernet (LAN) CI+ 3.5mm jack Optical digital audio output DVB-T2/C, DVB-S2 3x HDMI 2.1 (CEC ALLM VRR), eARC (HDMI 2) 1x USB 2.0 1x USB 3.0 Ethernet (LAN) CI+ 3.5mm jack Optical digital audio output power 230W Standby consumption: ≤ 0.5W 500W Standby consumption: ≤ 0.5W operating system Google TV Google TV SOUND Speakers: 2 x 15W Dolby Audio, DTS:X Speakers: 2 x 15W Dolby Atmos others Google Cast Google Assistant Apple AirPlay Google Cast Google Assistant Apple AirPlay price From 549 euros From 1,399 euros Two families with important differences The Xiaomi TV S Mini LED 2026 range is divided into two blocks differentiated by the diagonal from your screen. The 55, 65 and 75 inch models share the same features, with special emphasis on the gaming performance thanks to the mode Game Boostwhich allows the native refresh rate to be scaled from 60 Hz to 120 Hz, although it limits the resolution to 1440p. Xiaomi did not want to skimp on dimming zones for this panel, offering 308, 384 and 512 zones attenuation local respectively for each of their sizes. This count allows you to better control the light that reaches the LCD panel and reduce defects inherent in LED technology like blooming and halos that wash out the intensity of the blacks around a very bright point in the image. In the sound section of this range, the audio system has Dolby Audio and DTS:X certification, which is not the same as Dolby Atmos. The models of 85 and 98 inches are another story. Here Xiaomi has tightened the screws on almost all fronts: the dimming zones jump to 640 and 880 respectively, the native refresh rate rises to 144Hz (with the possibility of reaching 288Hz in Game Boost), the three HDMI ports are 2.1 with support for VRR and full ALLM that support 4K signal at 144Hz. To top it all off, these models are compatible with Dolby Vision and Dolby Atmos, which makes them much more serious options for enjoying home theater or major sporting events like the World Cup. The processor of this large-inch range also improves, going from the Quad Cortex A55 to the Quad Cortex A73, with 3 GB of RAM compared to the 2 GB of the smallest models in the range. In all cases the operating system is Google TV, with Google Cast, Google Assistant and Apple AirPlay integrated. Xiaomi lands on large-inch televisions If there is something that differentiates this generation from the previous ones, it is precisely Xiaomi’s determined commitment to large inch formats taking its range beyond 85 inches. For years, a 85 or 98 inch television with MiniLED technology It was territory reserved for premium models from brands such as Samsung, LG or Sony, and with prices that easily exceeded 3,000 or 4,000 euros. However, today the scale of manufacturing large-inch panels and the increased production of panels MiniLEDs allow Xiaomi to put a 85-inch QD-Mini LED TV at half price that just a couple of years ago. With this proposal, the Chinese brand puts the finger on the sore from its competition, putting pressure on the mid-range and lower-middle segment with models with very good performance and technological equipment at an unbeatable price. Also, put your foot in the big inch segment which, until now, was reserved for the top televisions of each brand and does so by pouring salt into the wound that hurts the most: that of price. Versions and prices of the Xiaomi TV S Mini LED 2026 The new Xiaomi TV S Mini LED 2026 range is available in sizes ranging from 55 inches to 98 inches and its prices are: Xiaomi TV S Mini LED 2026 55″: 549 … Read more

The next Mercedes-Benz model aims like a missile to fully enter the war

by

In the middle of World War II, while Allied bombing destroyed German factories and consumed resources at an impossible rate, many plants that until then manufactured cars, engines or civil machinery began to transform hurriedly to produce military vehicles, aviation parts and weapons. Some of the most recognizable brands in the European automotive industry they then discovered something that decades later resonates strongly again: in times of geopolitical tension, an assembly line can change purpose much faster than it seems. The unexpected twist, or almost. For decades, the future of the European automobile seemed to come down to a single discussion: electric, hybrid or gasoline. However, the German industrial crisis and the accelerated rearmament of Europe are opening a possibility completely different. Mercedes-Benz, like before Volkswagenhas just made it clear that it is willing to enter the defense industry if the business makes economic sense. This has been confirmed through an interview in the Wall Street Journal of its CEO, Ola Källenius, and it is much more important than it seems because it reflects a profound change within the German automobile industry: the big brands are no longer only looking at the car of the future, they are also beginning to look at war as a new industrial opportunity. In a Europe increasingly obsessed with drones, missiles, air defense and military production, car factories are beginning to be seen not only as car plants, but as possible centers strategic manufacturing. The perfect storm. The context explains why this idea is beginning to seem reasonable even for companies historically far from the military business. The German automobile industry is going through one of its most delicate moments in decades: falling profits, pressure from Chinese manufacturers, high energy costs, lower European demand and tariff threats from the United States. Mercedes-Benz, for example, suffered a strong profit drop in 2025, while practically all major German manufacturers have announced cuts or adjustments labor. At the same time, the defense industry is experiencing exactly the opposite situation. European rearmament after the war in Ukraine has fired orders, investments and military contracts to historic levels. For many German industrial companies, the military sector is beginning to represent something very different from a marginal business: stability, growth and guaranteed public financing for years. From cars to artillery. The case of Mercedes is not isolated and we have been counting. Volkswagen is also exploring possible military collaborations as defense companies such as Rheinmetall study reuse factories of automobiles or absorb part of its industrial infrastructure. The message is clear: Europe is beginning to discover that many capabilities necessary to produce modern cars (advanced metallurgy, electronics, robotics, complex logistics chains or highly skilled workers) are also extremely useful to manufacture systems military. The border between both industries begins to fade little by little. It is no longer just about producing tanks or ammunition, we are talking about radars, drones, autonomous vehicles, electronic systems and air defense platforms that require technologies very similar to those of the modern automobile. The new European war economy. As we said, the ukrainian war It has caused an enormous psychological change within Europe. For years, much of continental industry assumed that globalization and stability made a large military capacity of its own unnecessary. Now the opposite happens: European governments are increasing defense budgets at speeds not seen since the Cold War. This transformation is pushing traditionally civil companies to reconsider their role within the new geopolitical context. The CEO of Mercedes himself insist that any military activity would remain dwarfed by its core business, but at the same time recognizes something revealing: can become a growing and profitable niche. That is to say, the German automobile industry is beginning to assume that part of future European growth could come directly from rearmament. The car of the future may not be a car. If you like, the most striking thing of all is the symbolism of change. For a long time, the automotive debate revolved around batteries, autonomous driving and sustainability. Now, some of Europe’s most iconic companies are beginning to speak openly on anti-drone defensemilitary production or collaboration with weapons manufacturers. The idea that the next big European industrial business could be closer to war than sustainable mobility would have seemed absurd just a few years ago. However, the combination of economic crisis, Chinese competition and continental rearmament is slowly pushing giants like Mercedes-Benz itself into completely new and unexpected terrain. And that reveals the extent to which Europe is entering a stage where the economy, industry and security are beginning to mix more and more. Image | Nara, RawPixel, Julian Herzog In Xataka | Europe wants to make more weapons and faster. Your biggest obstacle is not money: it is finding qualified welders and technicians In Xataka | In the midst of rearmament, Spain has just surprised Europe: 5,000 million for 34 warships and four submarines

There is a battle to have the AI ​​model that programs best. And a good, pretty and very cheap rival has appeared in it: Cursor

by

Cursor has introduced Composer 2.5a generative AI model specifically intended for one thing: programming well. How good? Well, according to this startup, it does it as well as the best models of the moment, Claude Opus 4.7 and GPT 5.5, but it also does it for a lower cost. The challenge is striking not only because of what it means for Cursor, but because of how they have created that model: it turns out that it is based on a Chinese AI model. AI models specialized in one thing. While OpenAI and Anthropic try to develop general-purpose models—they do a lot of things really well— Cursor you have decided to focus on a specific task. The AI ​​startup has created an AI model specialized in programming, and has done so by arguing that a billion parameters are not necessary to compete with the best. Devoting yourself to a single thing allows you to not only gain efficiency, but also costs. This is not a decathlete, but a specialist in the 200 m event, so to speak. As good as GPT-5.5 or Claude Opus 4.7? That’s what they say in Cursor, because according to their tests with several specific programming benchmarks, the performance is on par with those two models that today are the great references both in programming and in other areas. And much cheaper. These results are also especially interesting when we add the cost factor. The average cost per task in the CursorBench 3.1 benchmark showed that Composer 2.5 managed to solve almost 65% of all tests for a cost of just $0.3. Opus 4.7 max and GPT-5.5 xhigh managed to reach that 65%, but at much higher costs: just over 4 dollars in the case of GPT, and 11 dollars in the case of Opus. The difference is abysmal. He API access price demonstrates the differences: 0.5 dollars per million input tokens 2.5 dollars per million output tokens, when Claude Opus 4.7 is 5/25 and that of GPT-5.5 is 5/30 respectively. Textual feedback. Unlike models that only learn from the final result, Composer 2.5 has been trained with a reinforcement learning technique (Reinforcement Learning) that allows us to offer clues about what is happening if errors are being made. This allows the model to recalibrate and act as a transparent teacher. One that also corrects word by word as it solves the exercise, not just when seeing the final result. 85% of the training budget has been dedicated exclusively to reinforcement learning, calibrating the model not for chat, but to execute code refactorings or fix bugs in real time. A model “born” in China. Those responsible for Cursor themselves have explained that Composer 2.5—like its predecessor, Composer 2launched at the end of March—is a model derived from Kimi K2.5, the AI ​​model of the Chinese startup Moonshot. Although that is the basis, already in Composer 2 the training and post-training tasks manage to improve the behavior in a very notable way in programming benchmarks and also in others such as Terminal Bench that evaluate the agentic behavior of these models. Cursor gets older. This startup became famous for creating a programming AI agent that was a pioneer in that fever we live for vibecoding. The user experience is no longer that of programming, as in traditional IDEs (Integrated Development Environments), but rather that of directing the machine to program it for you. Composer 2.5 doesn’t just program: it understands the structure and relationships between files, and turns Cursor into a much more competitive AI company, because it no longer depends on being able to work with Anthropic or OpenAI models, for example. Having both the AI ​​agent and the model processing everything makes it a much more competitive solution. Elon Musk has Cursor in his sights. Cursor’s good performance has led to growing interest in buying this company even before it becomes too big. Elon Musk knows this well and Grok, xAI’s model, is not so popular in the programming field. In April we learned that SpaceX had reached an agreement that gives you the option to buy Cursor for 60,000 million dollars. It would be a promising deal for both, because Composer 2.5 has already used Colossus’ infrastructure to train, and xAI could thus try to gain market share in the juicy enterprise sector. In Xataka | Elon Musk knows that TSMC is overwhelmed: Terafab is his idea to completely change the global chip industry

DeepSeek wants to raise its first round of financing and copies the last thing that remained to be copied from the US: the economic model

by

Chinese AI startups appear to have surrendered to Silicon Valley capitalism. Both DeepSeek such as Moonshot AI (Kimi) have begun to raise investment rounds or are preparing to do so. It is a turning point in a race that is now becoming especially interesting and that also raises a clear question: will these companies continue betting on open models? The valuation is multiplied by two. DeepSeek had always avoided making that decision and it seemed almost a personal project of its founder, billionaire Liang Wenfeng. However, the company is now in talks to raise its first round of external investment, they assure in Financial Times. According to company data, Wenfeng has 89.5% of the stake in the company. There is talk of a round that would increase DeepSeek’s valuation from the current $20 billion to around $45 billion. Who is the “Big Fund”. Behind this investment round is above all the China Integrated Circuit Industry Investment Fund, also known as the “Big Fund”. This consortium, the most important of its segment in the field of semiconductors, is supported by the Chinese state, and has a “cash” of 47 billion dollars contributed by the Chinese Ministry of Economy, the local government and several state banks thanks to a third round that was carried out in 2024. At the moment the “Big Fund” has not invested in other Chinese AI startups, but it has in companies like SMIC or Yangtze. The war for talent. The reason behind this decision is not only the need for capital to have access to more computing capacity. According to sources close to the operation, Liang Wenfeng has been forced to open that option to stop talent theft and thus be able to keep their best researchers on the payroll. In a market as competitive as this one, DeepSeek needs to offer shares to its employees to compete with the aggressive recruitment of talent by its local and Western rivals. A promising pairing. The relevance of this investment goes beyond the AI ​​model as such. DeepSeek has been significantly optimized for be able to run on Huawei hardwareallowing China to have a platform that works without the need for Nvidia chips. This symbiosis between this efficient AI model and the Chinese hardware giant is quite a bet by the Chinese government to try to win this race despite Washington’s blockades. The forced bet on “national” chips. Seeking that support in Huawei chips is not only a technical choice, but a political necessity for survive NVIDIA GPU crash. The problem is that Chinese hardware is still struggling to close the raw performance gap against architectures like Blackwell’s. If DeepSeek’s software hits a ceiling and chips created in China do not evolve at the necessary pace, the laboratory could find itself trapped: it would not matter to be very efficient when they cannot compete in raw power. Moonshot signs up for the rounds. DeepSeek is not alone in this race to achieve huge valuations. Moonshot AI just got up 2 billion dollars from investors such as Meituan, raising its value above 20 billion. Meanwhile, other rivals such as MiniMax and Zhipu AI (GLM) already surpass the 30,000 million valuation in their stock market debuts. This trend is therefore following what was already experienced (and continues to be experienced) in the US with AI startups, and the capital bubble that exists in the North American country now seems to have its eastern version in China. Moonshot AI and exceeds $200 million in annual recurring revenue (ARR). The paradox of copying the economic model. It’s ironic that DeepSeek, which became famous for challenging the “brute force” of American spending, ends up adopting its same funding structure. The company has shown that efficiency could offer an alternative to those almost unlimited resources of venture capital accessed by OpenAI or Anthropic. However, market reality dictates that a very solid capital structure is still what is needed to survive in the long term. Either you have it, or you can’t continue training models, reserving computing capacity and, of course, retaining talent. Open models? Until now DeepSeek had been one of the heroes of open weight AI models. Thanks to this, platforms like Hugging Face allow you to download it and allow everyone to take advantage of its achievements in terms of efficiency. The entry of venture capital and state funds could change the rules of the game: investors do not usually inject billions of dollars so that the product ends up being “given away” even for its competitors. The company will probably face the dilemma of closing its next models to protect its valuation and generate exclusive income, or keep its philosophy open at the risk that its investors no longer trust that strategy. In Xataka | If at some point NVIDIA has to choose between giving its best chips to the US or China, its choice is very clear.

Chrome has always liked to gobble up RAM. Now download a multi-gigabyte AI model without warning

by

Chrome is part of the digital routine of millions of people to the point that we often stop wondering what exactly it does while we browse. We use it for almost everything, we trust it with sessions, extensions, passwords, searches and a good part of our life on the Internet. That is why it is so surprising to find a folder larger than 4 GB associated with an AI model downloaded by the browser itself. We are not talking about a minor update or a residual file, but rather a large component that many users probably did not expect to see there. The conversation began to take shape from a publication by Alexander Hanff in That Privacy Guy. Their finding, in essence, was simple to understand: according to its logs, Chrome had left a multi-gigabyte AI model on his computer without giving him a clear warning during the process. From that clue I did the checking on my own computer, used from Spain, and found the same folder that Hanff refers to: OptGuideOnDeviceModel, within Chrome’s internal files. In my case, macOS shows that folder as 4.27 GB in size, even though features like the Gemini sidebar are not yet available in this market. Gemini Nano downloaded to my computer Gemini Nano It does not work like a traditional download that we search for, accept and install manually. In the Chrome developer documentationthe company explains that the integrated AI capabilities are intended to be fluid and that model management is done automatically in the background. It also notes that the initial download can be triggered when an AI feature built into the browser needs to use the Gemini Nano for the first time. In other words: the model can reach the computer as part of Chrome’s internal workings, not necessarily through a clear and recognizable action for the user. An AI model that goes beyond an integrated chatbot The model is not limited to promoting a browser with a chatbot integrated within Chrome. Google has already described uses Gemini Nano on the device itself to detect technical support scams, a type of threat that often lasts a very short time online and can escape traditional tracking systems. In that scenario, Chrome can provide the model with content from the page the user is visiting to extract risk cues. AI, therefore, can also be part of the browser’s security layer. Gemini Nano also boosts security features in Chrome That’s where a good part of the discomfort lies. AI in the browser can have reasonable uses, from helping detect fraud to powering writing, translation or summarization functions, but the problem arises when the user does not fully understand what has been downloaded, why it is there and how they can manage it. Hanff sums it up with a very direct criticism: “Chrome didn’t ask. Chrome does not show it to the user. If the user deletes it, Chrome downloads it again.” There are also voices that reduce the seriousness of the case. On Reddit, a user defended that the model is only downloaded when someone tries to use an AI function that needs it and that it can also be disabled from the Chrome options. Hanff responded that his logs showed otherwise: the browser opened on schedule, stayed on a page for a few minutes without interaction, and still left a trace of the download. Beyond that specific discussion, Google’s own documentation points to a middle ground: the download can be triggered by built-in functions and continue in the background even if the tab that started it is closed. Chrome does offer controls to reduce the presence of some AI features, but it doesn’t concentrate everything in a single, easy-to-understand panel. From settings can be disabled or hide certain visible pieces, such as Gemini in the markets where it is available, typing assistance, search history or AI-powered search. To go deeper, however, We must enter more technical terrain, such as experimental options from chrome://flags. This jump changes the experience quite a bit: we are no longer talking about turning off a clear function, but rather touching internal parts that may also be linked to features that the user may want to keep. Firefox offers an easy way to disable AI features Firefox offers an interesting counterpoint because Mozilla has grouped its AI controls in its own section within the settings. Since Firefox 148, that section is now available as “AI controls” and allows you to block current and future improvements from a visible place, without having to chase options spread throughout the browser. It also separates specific sections, such as on-device AI, translations and chatbot providers in the sidebar. It is a more direct approach: the user not only sees that these functions exist, they also better understand what they can activate, block or leave available. The arrival of Gemini Nano to Chrome is part of a broader movement: browsers want to become more than just a window to the Internet and start executing AI tasks within the computer itself. That direction can have real advantages, especially if it serves to strengthen security or make some functions more agile. But the case also leaves a visible panorama. Some users won’t mind at all that Chrome downloads local models automatically; others, instead, they will want to knowunderstand what it is for and have room to decide. Images | Xataka with Grok | Screenshot In Xataka | It doesn’t speak, it doesn’t climb stairs and it doesn’t even always obey: this is the robot that the creator of the Roomba has been wanting to develop for 30 years

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.