openai

GPT-5 is being a tortuous path for OpenAi and points to an uncomfortable reality: the era of the head is over

by

Chatgpt-5 is falling. They have not confirmed it directly, but Sam Altman is already in ‘Hype’ mode and all Rumors They point out that the new model will be presented this month. That GPT-5 It will improve its predecessors is evident. However, as has happened in other industries (Did anyone say smartphones?) The rhythm of the advances is slowing down and everything indicates that GPT-5 will not be the revolution that were previous versions. A discreet jump. They tell it in The Information. Sources close to OpenAI say that, although with improvements, the qualitative leap will not be as great as it was with GPT-3 and GPT-4. This contrasts with Altman’s constant mentions The close thing that is to achieve a general artificial intelligence (AGI), an AI capable of matching the smartest human being in all areas of knowledge. But Chatgpt-5 is far from being AGI and also OpenAi has encountered numerous obstacles along the way. The GPT-5 that was not. At the end of 2024, Openai was working on an internally nicknamed model and that was the candidate to become GPT-5. However, the jump with respect to the previous version was not large enough and finally OpenAi decided to launch it as GPT-4.5, an intermediate version that Nor was it a great improvement With respect to its competition. In Xataka A group of experts in AI attended a party in a mansion. The topic of conversation: what will be when AI ends humanity The one that will be. Openai has continued working on what GPT-5 will be and is about to launch. According to internal sources, the new model improves especially in programming and also does so more efficiently; It is able to produce higher quality code without consuming more computing power. Openai wanted to improve this area, where Anthropic takes a lot of advantage with Claude Code. GPT-5 will also improve its agricultural skills and will be able to follow complex instructions to function without human supervision. In The Information they cite an example about a customer service agent. Giving it a list of rules, GPT-5 can decide when a client must receive a refund, while in previous models they needed to see many complex cases to do it well. Although there will be improvements, The shadow that will not be surprising continues to fly over. Obstacles. The Tensions between Openai and Microsoft They have hindered the advance of the new model. There have been internal friction between both companies, especially for intellectual property issues. Openai accused Microsoft of anti -competitive behaviors. Microsoft accused Openai of not giving access to his most advanced models. They also faced a shortage problem of quality training data and, more recently, the Discompassed talents to the team of Zuckerberg Superintelligence and Google by removing Windsurfhis commitment to the Vibe-Coding. {“Videid”: “X9N4GWC”, “Autoplay”: False, “Title”: “OpenAi has presented chatgpt agent”, “Tag”: “chatgpt”, “duration”: “47”} The challenge. Openai has encountered a problem: that specialized reasoning can be taken to a chat with which we can talk naturally. It is what happened to them with O3, the reasoning model that They presented in April and that will feed GPT-5. Both this model and O1 were trained by reinforced learning, but O3 provided much more processing capacity and obtained surprising results. However, by turning it into a version of chat, that performance fell into chopped and I lost many capabilities. The reasoning model is so powerful that the response time increases a loteven for very simple questions. GPT-5 is expected to be at an intermediate point. Continuism. GPT-5 improvements may be more than enough for OpenAI continues to feed your machinerybut it seems clearer that there is a deceleration of the generative AI. At the end of 2023 Bill Gates already said that I did not believe that GPT-5 would be much better than GPT-4 and that the generative AI was close to its roof. It is an almost natural process and that we have seen with other industries such as smartphones: a lot of innovation and qualitative leaps in the first years, followed by a continuous stage in which the novelties are lower and the sensation is progress, but much slower. Image | Wikipedia, Pexels In Xataka | There is a chatgpt fever among public officials. What we do not know is how it will affect us as users (Function () {Window._js_modules = Window._js_modules || {}; var headelement = document.getelegsbytagname (‘head’) (0); if (_js_modules.instagram) {var instagramscript = Document.Createlement (‘script’); }}) (); – The news GPT-5 is being a tortuous path for OpenAi and points to an uncomfortable reality: the era of the head is over It was originally posted in Xataka by Amparo Babyloni .

OpenAI is already generating GDP size benefits from a small country. Follow light years of being profitable

by

Winning 12,000 million dollars a year seems somewhat prodigious for any company, but not when that company is called Openai. The evolution of income is being remarkablewithout a doubt, but both her and others – and here Anthropic is another good example – something serious happens to them: that they continue to spend more than they win. 12,000 million for OpenAi in 2025. As indicated In The Informationa new estimate that Openai’s “annualized” revenues will be 12,000 million dollars in 2025. The figure is a projection, but it is significant taking into account that in 2024 the estimated revenues were according to various sources of 3.7 billion dollars, although In Reuters They talked about the fact that they had actually reached 5.5 billion dollars. And 4,000 for Anthropic. The same media also recently indicated how the estimate in the case of one of its great rivals, Anthropic, has also risen and now It is 4,000 million dollars. Just two months ago that figure had already been checked and was 3,000 million, which means one thing: both are growing in number of subscribers. 700 million “Chatgpteros”. Another of the data to which the information article points out refers to the number of weekly active users. According to their data, 700 million people use chatgpt at some point in the week, which marks a unique milestone for the company. It is true that the vast majority of them are users of the free version, but that base is what allows part of those who use the service for free They end up pointing out to any of the chatgpt subscription plans. Income growth is being unusual in OpenAi and Anthropic, but both companies are spending absolute fortunes to end up being profitable. Source: Reddit. It will win 12,000 million, how much will it spend? In Reuters indicate that the internal estimates of the company also point to higher expenses. According to those projections, OpenAI will spend 8,000 million dollars, but that figure is dentra on direct operational expenses. There are many more associated expenses – investments, infrastructure, other financial obligations – and that makes OpenAi not profitable for now. We do not have estimated spending data for Anthropic, but it has an identical problem: Spend more than you earn. Spectacular, but. Although this growth in income is certainly extraordinary, it must be taken into account that to achieve this, these companies carry “Burning money” for years. The investment rounds that Anthropic and especially OpenAi have captured have allowed them to have a lot of room for maneuver to lose huge amounts of money without that at the moment that worries too much. And they will continue to spend as possess. Especially in the case of Openai, which thanks to SoftBank support It has great plans that will make it necessary to spend true fortunes. They have done it to Buy the Jony Ive design study for 6,500 million dollars, but above all they will do it with the project Stargatewhich still seems like very difficult to complete. But no profitability until 2029. Those responsible for OpenAi do not seem too worried, and we knew what the company’s financial road map was known weeks. They will continue losing money until 2029when supposedly – all is a free estimate, not a promise – will earn 100,000 million dollars. It will be then when the company will begin to be really profitable, but again, All this is a promise (or maybe a hope). It could not perfectly be fulfilled … and even ending up falling short. Image | das | Fortune Brainstorm Tech In Xataka | Chatgpt takes the step to conquer students and teachers: their new mode does not give the answer, I build it with you

Openai has invested billions in chips. His only problem is that he has forgotten something important: cables

by

While Openai designs increasingly complex models and promises cities size data centers, a much more terrestrial reality threatens to stop the advance: the electrical infrastructure is not ready for what is coming. At the beginning of the year. Sam Altman (OpenAI CEO) promised, together with Oracle and SoftBank, an investment of 500,000 million dollars in AI infrastructure Under the name Stargate. The announcement, Made at the White House with President Trumphe talked about building 10 gigawatts of data centers on American soil by 2029. However, the start has been slower than expected. According to The Wall Street Journalnot a single formal contract with SoftBank has not been signed, and the first center, in Ohio, is still in the evaluation phase. Meanwhile, Openai has advanced on his own, expanding his alliance with Oracle to develop 4.5 more gigawatts, adding more than 5 GW under construction, As the company itself has reported. A big problem behind. This does not only affect Openai, rather the entire AI sector since no investment or quantity of GPUS can solve the true bottleneck of the sector alone: the electricity grid. As Le Monde explainedtraining models such as GPT-4 consumes dozens of gigawatts -hora, but the real challenge is in the “inference”, that is, in daily use. Each consultation to a model like Chatgpt implies complex calculations that consume energy every second. From the International Energy Agency (IEA), It was warned That the global electricity consumption of data centers could double before 2030, exceeding 945 TWH, more than all Japan today. And there the paradox appears: we can generate more energy, but we have no how to move it. In other words, high voltage lines, transformers, substations, adequate land, permits, technicians are missing. It’s like wanting to fill a city with bottled water, but without having pipes. The project continues. In a last statementOpenai has affirmed that its expansion with Oracle is already creating tens of thousands of jobs in Texas and that the objective of the 10 gigawatts is on its way to being surpassed thanks to new alliances. The Stargate I site, in Abilene, has already begun to operate with NVIDIA GB200 chips. However, As the Wall Street Journal has detailed, Disagreements with SoftBank persist on where to build, how to finance and how to connect data centers to electrical networks that are already saturated. Sam Altman recognized the challenge in an internal memorandum cited by axios: “The thirst for computer science is beginning to tension the supply chain and demands some real creativity.” Other paths: atomic energy. Faced with these limitations, technological giants are looking for solutions outside the traditional electrical system. The answer, for many, is in a surprising return: nuclear energy. Goal has signed a 20 -year contract with Constellation Energy to supply part of its data centers from a nuclear plant in Illinois. Google and Amazon They have also opted for small modular reactors (SMR). Microsoft, meanwhile, will reopen a closed nuclear power plant since 2019exclusively to support your AI infrastructure. Nvidia has not been left behind. In 2024, it invested 650 million dollars in Terrapower, the company founded by Bill Gates that is building the first Natriat reactor, a fourth generation machine that, According to their developersIt will generate electricity by half of the cost of a conventional reactor. The project, which takes place in Wyoming, has a Spanish participation: the public company sees the reactor cover. Without cables, there is no ia. Meanwhile, the number of users continues to grow. Chatgpt reached 800 million active users in April, According to Altman cited in axios. Each of them generates requests, questions, images, instructions, and all that consumes energy. According to Le Mondetasks as simple as writing an email can be more than 7% of a complete mobile phone charge. And generating an image consumes even more. Elon Musk says his company XAI already operates with 230,000 GPUS, and expects 550,000 more. The AI does not rest, but the cables that feed do not supply. Artificial intelligence promises to change the world, but before it will have to face something more basic than any algorithm: the physical laws of electrical systems. There is no needless. There are no servers without energy. And there is no energy without cables. Image | Pexels Xataka | The AI is opening the doors of a radical revolution on the Internet: that we can all create apps without knowing

Openai and Oracle prepare a brutal data center with 2 million chips for ia

by

The plan ‘Stargate’ It is the great bet of the government led by Donald Trump to keep the US at the head of development in artificial intelligence (AI). When this project was made public on January 22, it hit the economic endowment that would make it possible: nothing less than 500,000 million dollars. This money will leave the coffers of the Japanese investment group SoftBank; of those of OpenAI, the creators of Chatgpt; of those of Oracle, and, finally, it will also be provided by the investment firm Emiraratí MGX. These companies will support the construction during the next four years of an advanced network of data centers that will house the high performance computing infrastructure necessary to sustain US leadership in the AI field. The spearhead of these facilities is already being built in Texas (USA), in a town called Abilene. And it is colossal. In fact, this first data center of the ‘Stargate’ project will bring together, According to OpenAimore than two million chips for ia. 4.5 GW Dan for a lot When the US government announced to Bombo y S pay this plan left a great question open: how did he plan to solve the supply of electricity required by the new facilities? Large data centers for AI consume a lot of electricity, which has caused Some technology have opted for investing in nuclear centrals to guarantee the supply of electricity that these facilities require. Oracle began to install the first ‘racks’ with NVIDIA B200 platform servers in June At the moment this question is not completely resolved. And it is not because the ‘Stargate’ infrastructure should be completely ready before President Trump’s current mandate expires, and a new nuclear power plant can hardly come into operation in four years. Even so, Openai and Oracle have formalized A few hours ago they have reached an agreement to build the necessary infrastructure to Deliver additional 4.5 GW to your data centers. Interestingly, SoftBank does not participate in the financing of this expansion, although, as I mentioned a few lines above, it does in the ‘Stargate’ project. According to OpenAia part of the gigantic Abilene data center is already in operation. In fact, Oracle began installing the first racks With platform servers B200 of NVIDIA In June. It is surprising that in just six months a part of this installation is already active, but it is important that we do not overlook that the Abilene data center seeks to demonstrate that the ‘Stargate’ plan is viable within the deadline that those responsible have promised. Anyway, in this equation there is another unknown that also has a lot to say: China. “We hope that China significantly increase its investments in AI and semiconductors in response to the US domain in AI,” CBM consultancy analysts foresee. It makes sense. These two great powers world supremacy are being disputedso it is understandable that each significant step that give one of the two Receive a more or less overwhelming answer from the other. We can be sure that 2025 will be an even more agitated year than 2024 in the geopolitical and technological fields, so we will be attentive to the steps that US and China will surely give. Image | Christina Morillo More information | OpenAI In Xataka | Huawei attacks Nvidia positions in China: he wants to have dominant hardware in inference processes in AI

We believed that Chatgpt was just a very capable chatbot. Openai has just turned it into something very different: a real agent

by

We have been talking about artificial intelligence agents for a long time, but Openai has just converted that conversation into something much more tangible. The company has presented Chatgpt Agent, a function that turns its popular assistant into something more autonomous: now it is able to execute complex tasks using a virtual computer, with tools that allow you to navigate, program or even make decisions. From Agent Operator. At the beginning of the year it presented Operator, a tool that allowed ChatgPT to interact with web pages. Then Deep Research arrived, focused on writing long reports from multiple sources. The background idea was clear: go beyond the conversation and approach real tasks. What has been presented today is something like a tool that unifies all these previous advances. During the demonstration, those responsible for the project raised a daily situation: organizing a trip to attend a wedding. The agent was able to understand the context, find hotels, propose gifts, take into account the weather, the clothing code and even remember that a suit had to be bought. He did it by analyzing the message, accessing the web and acting step by step, as a person would. The difference is that everything happened within Chatgpt, without the need to alternate tabs or give instructions one to one. A virtual computer for AI. The key is that the agent is not limited to responding to text: it operates within a kind of virtual computer that Openai has given access. You can use a text browser to read pages quickly, a visual browser to interact with buttons and forms, and even a terminal to run commands, generate code and manipulate files. You can also work with spreadsheets, presentations, and access services such as Google Drive, Calendar or Github if the user authorizes it. What is under the hood? The model that drives chatgpt agent (specifically developed for this function, although without official name) was trained with complex tasks that required to combine multiple tools. Openai used reinforcement learning, the same approach that you already use in its reasoning models, to teach you to choose when to use the browser, the terminal or an API. The idea was to develop a solution capable of accurately deciding how to act based on each objective. In development. Images | OpenAI In Xataka | Goal is in a hurry to lead the AI that has done something unusual: it is building a data center in tents

This is the goal plan that has stolen brains to OpenAi and Google

by

Mark Zuckerberg is distributing checks as if there were no tomorrow. The Meta CEO announced yesterday a great restructuring of its artificial intelligence division, which now aims to “superintelligence.” And to reach it, Zuckerberg is stealing engineers from OpenAi, Anthropic and Google with a carver. Restructuring with two star signings. The new division of the company is called the SuperintenTintelligence Labs (MSL) goal, and will be led by Alexandr Wang, Excel of Scale AI. That has been the first star signing of Zuckerberg, which decided to invest 14.3 billion dollars in that company to buy an important participation (49%) but above all to get Wang’s services. The second great signing is Nat Friedman, Exceiver of githubwhich will co -direction that new division with Wang and will focus according to the statement on AI products and also in R&D. The internal statement, to which it has had access Bloomberg And that also shares CNBCreveals more interesting details. Up to 100 million bond. To form this new team, Meta has offered bonds of up to 100 million dollars to the most important engineers of other companies. The salaries and compensation offered in detail are not known, but the figures are simply astronomical. In Bloomberg they point out that the company has offered some investigators shares worth dozens of millions of dollars. Stealing employees to the greats of AI. The engineers who have managed to sign a goal They come from OpenAIGoogle, Deepmind and Anthropic. These are experts who have been the great ones responsible for developing and training products such as GPT-4O, Gemini 2.5 or Claude, and precisely that experience is the one that goal wants to take advantage of making a definitive qualitative leap when it comes to dominating the market. Media like Wired have A list With 11 of those star engineers. And Lecun, what? In that internal statement, however, There is a name that is not mentioned and that is especially relevant. We talk about Yann Lecunthat until now had been visible head of the efforts of the finish line. Lecun has been making it clear that the generative AI It is not the valid path To achieve artificial superintelligence or an AGI. His speech It has always been almost “anti -judicial”, and has been Very critical With the ability of current models, although it has always supported the launch of IA Open Source models, of which Llama is the greatest exponent. Better spend a fortune than regret it later. According to Bloomberg Zuckerberg, he stated that Meta will spend “hundreds of billions of dollars” in AI projects in the coming years. Last summer He already indicated that “it is very likely that many companies are oversizing (their investments in AI). But, on the other hand, I think that all the companies that are investing are making a rational decision, because the disadvantage of being left behind is that you could stay out with the most important technology of the next 10 to 15 years.” And there are also the data centers. Although there are no specific figures on the amount of money that has been spent on salaries and compensation for all that new division, the single investment in Scale AI (14.3 billion) and those hypothetical bonds of tens of millions of dollars already make clear the dimension of the bet. But that’s not all: Mark Zuckerberg himself already indicated months ago that this 2025 plan to invest 65,000 million dollars In data centers. It is a “shy” figure in front of the 75,000 What does Google plan to invest, 80,000 from Microsoft or the 100,000 Millions of dollars that Amazon estimates that it will invest in these facilities for training and inference of AI. Call on the one hand, new models for another. Zuckerberg also explained in the statement how he has a lot of confidence in the progress of flame 4.1 and calls 4.2, which are the goal base AI. However, he says that “in parallel, we will begin to investigate in our next generation of models.” But. The truth is that today calls is an absolute reference for researchers and startups that create their own projects from it, but its performance is behind the most capable models today. Meta has not achieved that the integration of Goal AI in WhatsApp or Instagram I just turned it into a preferential option for AI users, and Chatgpt remains the clear clear in this area. One thing is money and another mission. This “galactic” team can have a problem as Chon Tang explained, of the Berkeley Skydeck Investment Fund, having billionaire salaries can be counterproductive in the Meta Mission In Xataka | The cover of the AI ​​is a goal. He has been betting on her for more than a decade and has much more than she calls

Years ago the “Paypal Mafia” redefined our vision of the Internet. Today the “OpenAi Mafia” points to the same with the AI

by

When Ebay bought PayPal In 2002 for 1.5 billion dollars, there was an internal earthquake. In the four years following the acquisition, only 12 of the company’s original employees were left. Paypal was a boiling talent, Entrepreneurship And brilliant people, and who “Ebay escaped” They ended up redefining the technological world. Now we are attending a similar phenomenon, but with another company: OpenAi. PayPal Mafia. This was baptized by the group of bright ex -employed PayPal who left the company to found your own startups. These projects ended up becoming true business and technological successes, and showed that in PayPal there was a lot of talent waiting for their own opportunity to undertake. The most outstanding examples of that mafia Paypal already make clear the dimension of that phenomenon: Elon Musk: The clearest example of how far the members of the PayPal Mafia arrived. He founded or co -founded SpacexOpenAi, Neuralinkand The Boring Company, in addition to buying a majority participation in Tesla and Acquire Twitter (Now x). Peterthiel: is the president of Palantirhe created the Founders Fund Investment Fund – which would later attract former partners like Ken Howery or Luke Nosek – and was the first Facebook external investor. Steve Chen, Chad Hurley and Jawed Karim: these three former PayPal employees YouTube co -founded Reid Hoffman: He founded LinkedIn Russel Simmons and Jeremy Stoppelman: Yelp co -founded Yishan Wong: He worked on Facebook and became CEO of Reddit for a season David O. Sacks: Founded Geni.com and Yammer They redefined the Internet. These engineers and entrepreneurs were a fundamental part of the transition to an internet focused on consumers, something surprising especially after the bubble of the Puntocom in 2001. Rarely in history there has been a phenomenon in former employees of the same company end up creating one or several companies even more successful than the one they left, but there is a previously relevant previous example. The eight traitors. In 1957 eight engineers, fed up with his boss in Seticonductor Laboratory Shockley, decided to lite the blanket and create their own semiconductor company. They called her Fairchild Semiconductor Corp, and those employees They went down in history like the eight traitors. Fairchild had a huge success, but ten years later his engineers also looked for new roads. Three of them –Gordon MooreRobert Noyce and Andy Grove – ended up joining in 1968 to create a legendary company: Intel. The other Paypal Mafia is the OpenAi mafia. Those phenomena seem to be now having a new iteration, but this time the company of which all this new movement is starting is OpenAi. This company has become absolute reference in the field of AI, but on the way to achieve this there has been a true internal war and many conflicts between several of the engineers and their CEO, Sam Altman. Those internal tensions were the ones caused the dismissal and later return to the company of Altman In November 2023, but the company was not the same since then. OPENAI MAFIA members. Actually the divisions that occurred in Openai began very soon, almost shortly after the startup were formed. In fact, the first member of that “Openai Mafia” was also from the Paypal Mafia: It is none other than Elon Musk. It is a clear example, but not the only one, and the members of that OpenAi mafia are the following: Elon Musk: OpenAi co -founder that ended declaring war on the company and Sam Altman. Co -founded XAI Dario and Daniela Amodei: these brothers co -founded Anthropicthe company that develops the family of Ia Claude models. John Schulman: After being one of Openai’s main researchers, he left the company to join Anthropic. Others former investigators and engineers Openai like Tom Brown, Jared Kaplan, or Sam McCandlish also joined the company later. Jan Leike: This researcher who worked in Deepmind and OpenAi He left this company and joined Anthropic in May 2024. His work focuses on the development of a “safe” (“super -align”) and scalable. Ilya Sutskever: Openai co -founded and it was her Chief Scientist, but in 2024 left the company To create your own startup, Safe Superintelligence. Look Murati: Openai’s exco also decided leave the company something after Sutskever, and just Create Thinking Machinesanother startup of AI. Bob McGrew, who also left Openai recently, have joined that project. Josh Tobin and Vicki Cheung: These engineers also came out of Openai to found Gantry, another startup of ia. David Luan: After being one of the top engineering responsible for OpenAi, he ended up co -founding Adept AI Labs. Andrej Karpahty: After working on Tesla’s autonomous driving systems and then in OpenAi, this expert and disseminator ended up founding Eureka Labs. Many promises, few realities. The fundamental difference between these two “mafias” is that while in the PayPal Mafia those companies that founded its creators were absolute balls, the situation of those startups that Openai exingenier have created is an absolute unknown. Safe Superintelligence Ya raised 1,000 million dollars in 2024 and it is rumored that Thinking Machines soon It will lift 2 billion of dollars. None of them have launched any product for the moment, and in these new companies there is a lot of expectation – and, talent is supposed – but little else. The future will say whether the impact of these former OpenAi employees is comparable to that of PayPal’s ex -employees. Image | OpenAI In Xataka | There are more and more billionaires putting their money in human rejuvenation: the last Sam Altman

A new future for Openai is opening. And Microsoft is increasingly out of it

by

Things are not going well between Microsoft and OpenAi. What in 2019 began as a promising idyll Waters six years later, and both companies have been looking for their own way. The break is not definitive, but that process begins to be delicate and is making them begin to get the dirty rags of the relationship. Growing tension. In The Wall Street Journal They point how “tensions are reaching a boiling point.” One of the factors that contributes to the current situation is Openai’s intention to become a profit company. The current existing agreement that defines Microsoft’s participation in OpenAI is a reason for dispute. Microsoft wants more. The central debate is what participation Microsoft will have if Openai becomes a profit company. According to the WSJ, Microsoft is asking for a greater participation than Openai is willing to give it. If OpenAi does not complete your conversion before the end of the year, It takes a risk to lose 20,000 million dollars. Dangerous accusations. According to sources close to these companies, Openai managers They have accused Microsoft of anticompetitive behavior during his business alliance. Thus, while in Openai they claim that Microsoft did not offer all the computing capacity he could, In Microsoft they protested Because Openai did not give access to their most advanced models quick enough. New girlfriend. Openai soon began to look alternatives and new brides. He found one that has in fact promised much more than Microsoft gave Openai. It’s about softbankwhich will theoretically invest 40,000 million dollars in Sam Altman’s company and will contribute to Gigantic Stargate Project. On the one hand they say one thing … In a joint statement cited in WSJ, those responsible for Microsoft and Openai pointed out that “we have a long -term productive alliance that has offered incredible AI tools for all. Negotiations continue their course and we are optimistic when continuing to create things together in the future.” … but on the other they do another. OpenAI intends to close the acquisition of Windsurf for 3,000 million dollars. However, the agreement with OpenAI indicates that Microsoft has access to all the intellectual property of Openai. And the problem is that the latter does not want Microsoft to have access to Windsurf’s intellectual property. They already have Github Copilot, and it seems that in Openai they are not willing to make Microsoft take advantage of that acquisition to improve their own platform. And meanwhile, they compete more and more. Microsoft has access to OpenAi models, but also Ear money Every time someone uses chatgpt. Sam Altman’s company continues to benefit from the gigantic infrastructure of Azure and his computing capacity. And yet, both have been taking steps to “decouple”. Microsoft continues to work In its own AI models —Phi 4 continues to progress – for example, and OpenAi is still looking for new partners And he is drilling Even with Google to also be able to take advantage of its cloud infrastructure. In Xataka | The Microsoft Agreement with OpenAI was just a first step. What they want (probably) is to get rid of them

OpenAI has cut the price of O3 80%. And as a gift to the sector has sent a price war that will destroy margins

by

Openai has moved token: the same day as presented O3 pro The price of O3 cut 80 %. It may only be segmentation, but very possibly it is also an advance of the price war that starts when the rivals already touch their performance and internal efficiency allows us to serve Tokens for cents. The new step is like this (prices per million Tokens): ENTRANCE EXIT OpenAi O3 $ 2 (before $ 10) $ 8 (before $ 40) OPENAI O3 PRO $ 20 $ 80 Gemini 2.5 Pro $ 1.25 $ 10 Claude 3.7 Sonnet $ 3 $ 15 Deepseek R.1-0528 $ 0.55 $ 2.19 That price narrowing comes because quality already converges in practice: almost all models “reason” comparable in common tasks, and hardware and inference efficiency improvements (on the side of OpenAi and the rest) reduce the marginal cost of serving each Token. The result is manual: The price becomes the main lever and segmentation appears as a containment dike —How the creation of O3 Pro, designed for customers willing to pay a lot due to low latencies, huge contexts or formal audits. For developers this is fertile land. 2 dollars a million tokens, AI ceases to be a luxury expense and looks like a CDN: Payment for use, without permanence contract. Yes tomorrow Gemini either Deepseek They offer better value for money, change is reasonably simple in infrastructure and Prompts. The blocking effect is transferred to the layer that surrounds the model (tools, data, user experience), not the model itself. At the contrary end of the cost spectrum, Apple has integrated its foundational models directly into operating systems. Any app can be invoked at home, free and without counters from Tokens No invoices at the end of the month: AI becomes as invisible as GPS or camera. It is another function. The play Remember the emergence of WordPress in the early 2000s. At that time, publishing online required or a lot of knowledge or a lot of money to pay someone to have it. WordPress loaded that barrier and democratized the publication. Twenty years later we gave the xataka legend to its creator precisely because of that. And now what? Possibly, an adjustment downward the margins in the base models and a migration of the business towards higher level services: observability, management of the management Fine Tuningprivate data accommodation, vertical agents … exactly what happened with the cloud when the price of the gigabyte sank and the value moved to the management layers. The “magic” of AI is blurred, but its usefulness – now much more affordable – opens the door to a much wider market. Just what Openai needs. In Xataka | O4-mini is much more than another model of AI. It is the Tesla Model 3 of OpenAI Outstanding image | Ilgmyzin in Unspash

Goal is so desperate that it is beginning to offer up to $ 100 million to investigators of OpenAi and Google

by

Goal is offering salaries between 10 and 100 million dollars to Openai star researchers, Google and other companies to hire 50 experts who lead their new superintelligence laboratory, for which Zuckerberg has entered ‘FOUNDER MODE ‘, as published by The New York Times. Why is it important. The Zuckerberg company has lost ground in the AI ​​race after Some stumbling blocks With her models she calls and the key talent escape, including the director of Research of IA Joelle Pineau, whom we could interview a year ago. Now try to recover based on a talonario. In figures. Meta offers reach nine digits per investigator, that is, 100 million dollars. Although they are not structured as a check for that amount. The packages go “seven to nine figures” according to Own sources of The New York Times. Following the typical goal model, it is likely that most be in actions (RSUS) that are distributed for four years. The actions are distributed quarterly for four years. In any case, it is a conjecture based on what was seen above. And these are unpublished amounts, well above the 2 million annually that were already considered exceptional offers. In addition, the company plans to invest 15,000 million dollars to buy approximately half of Scale AI and bring to its CEO, Alexandr Wang, 28. The context. Meta created his first AI laboratory in 2013, but since the launch of Chatgpt in 2022 there has been somewhat lagging behind. His latest models have had performance problems, and the company was discovered after manipulating Benchmarks to make their products seem better than they really were. A practice that We saw in the past on smartphones and that also reached AI. Between the lines. Wang’s choice is not accidental: he is Zuckerberg’s personal friend and a billionaire who made his fortune with Scale AI, a company specialized in labeling data To train AI systems. Its closeness to political power could also be an aid at a complicated regulatory moment for the goal. Google, Openai and Anthropic continue to launch increasingly powerful models while finishing struggle to stay relevant. Sam Altman continues to insist that we are close to the AGI. Even more pressure. Yes, but. Money does not guarantee success. Goal is reputable to be A complicated environmentwith constant internal struggles. In addition, many of the best researchers already have millionaire offers of their current employers and may prefer to stay where they are. Deepen. This movement is part of an upward trend: large technological ones are buying promising startup parts to get talent without having to get the complete company. Microsoft He did it with inflection ai and Google with Character.AI. It is a way to save a few dollars, but above all, to dodge compromising questions by regulators. The big question is if Zuckerberg, who already burned a lot of money with a metaverso that remains far from the expectations raised (and the return of investment), this time can be successful with its most expensive bet to date. In Xataka | Goal has fired 35,000 workers in five years. And many of them fear having entered their “black lists” Outstanding image | Goal

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.