nAct as an expert news reporters or journalists and create deeply engaging, well-researched, plagiarism-free news article BASED ONLY AND EXCLUSEVELY ON INFORMATION FROM TEH ARTICLE BELOW, utilizing web search for relevant information and hyperlinking all external references directly to the contextual keywords within the blog body (NOT IN footnotes or a separate references section), including all provided quotes verbatim in quotation marks and attributing them naturally, seamlessly incorporating all multimedia elements from the original source, maintaining a sophisticated yet conversational tone with varied sentence lengths, integrating primary and secondary keywords organically, embedding relevant internal and external links, adding one table to summarize key points, strategically placing calls to action, fostering user engagement thru fresh insights and meaningful analysis, and returning only the requested content without any additional commentary or text.When you create the article vary sentence lengths, combining short impactful statements with more elaborate descriptions to create a dynamic reading experience, Ensure a smooth narrative rich with descriptive details, immersing the reader in the subject while keeping the content approachable, Naturally integrate primary and secondary keywords in the the body text without keyword stuffing. Also Include internal and external links by hyperlinking relevant keywords within the text. all backlinks must be hyperlinked directly in the body of the blog, not in footnotes or a separate references section.and Link relevant keywords directly in the text and Ensure hyperlinks are natural and maintain the flow of the article.
Do not place the sources at the end of the blog.YOU MUST HYPERLINK TO THE CONTEXTUAL WORD THROUGH OUT THE BLOG.Include one table in the blog post to summarize key information or comparisons, helping break up the text and present data in a digestible format and Vary Sentence Length: Mix short and long sentences to create a more natural flow and Be mindful of overusing certain terms or phrases, as this can signal AI authorship.
Do not place the sources at the end of the blog.YOU MUST HYPERLINK TO THE CONTEXTUAL WORD THROUGH OUT THE BLOG.Return only the content requested,without any additional comments or text.
The created article should be BASED ONLY AND EXCLUSEVELY ON INFORMATION FROM THE ARTICLE BELOW :nn:rnrn
The models revolution of artificial intelligence (AI) of China Depseek is based on a key motive: they do the same but cheaper. Its general operation is equally good that Silicon Valley market leaders, but are much less expensive to create and use. Openai, Anthropic and Google, creators of the moast famous models, and Nvidia, in charge of the sophisticated chips that use these companies, have seen all their apparent advantage in just a few days have collapsed.Today there are fewer experts who believe that training and using an AI model requires the power of a technological giant. The Monumental drop in Nvidia Stock Exchange It is the illustration of that belief.
Los chatbots de la china DeepSeek They work the same as everyone. The user is discharged with an email and a password on the company’s page and can start asking what he wants. Sometimes it takes more to answer, it does not always answer delicate political questions about Tiananmen or taiwan and works quite well in Spanish.The best thing, however, is that it is free. Deepseek’s business model is based on charging users who require professional use.
R1 is the most recent model of the Chinese company, which has caused the collapse of the Western AI building. Reason just like openai’s most advanced creation, O1. But in full Christmas holidays, the Chinese launched their V3 model, equivalent to the most common ChatGPT-4, which already included the technical advances that have generated the stir.
Deepseek has managed to train a new model, which cost more than $ 100 million in the case of ChatgPT-4, has passed to less than 6 million. Andrei Karpathy, co -founder of Openai, former director of AI in Tesla and one of the most respected experts in the sector, described that budget as “joke” and added: “We must ensure that we are not wasteful with what we have and this model seems like A good demonstration that there is much to review both in data and algorithms. ” The latest Deepseek model is so efficient that it required a tenth of the computer power of the comparable goal model.
These AI models are extremely expensive both to train and after using. Deepseek has managed to lower costs in both cases. To train them, he has made his model learn from previous others. This is something that the terms of service of these companies do not allow, but it is very difficult to detect and avoid. Like Openai’s original models they took all Internet data without paying for them, now a Chinese model uses closed models with a minimum cost. This is achieved with a technique called distillation, which allows you to ask the previous model and save the answers and thus learn and train. This technique implies that models can learn from each other, which throws us to another scenario of even faster and faster evolution.
the other technique that Useek uses to reduce costs is a system called “experts.” Every time the user asks a question, the model decides if he must activate his medical expert, translator, lawyer or physical. Classic models activate everyone suddenly, which is a waste of energy and computing.It is like needing a huge electricity plant to launch an elaborate assembly chain. Deepseek has done the same connecting his small factory to a plug and a handful of generators.
Suddenly, users who so far pay about 20 euros for using OpenAi models see how they can get almost the same free results. The expansion of AI is arriving faster than expected to its stage of comfortingwhen the new technology becomes a common product and accessible to all.
All this has also happened in a company in Hangzhou, a Chinese city near Shanghai. The Biden administration vetoed the export of the newest chips to limit Chinese competition. the response of the Asian giant has been simple: we do not need them. And they have succeeded with an open source, teaching the guts of their achievements with documentation published on the Internet. Make the need virtue.
The end of this initial stage of AI seems to be bad news for pioneer giants, who see how a good part of their advantage disappears. But the rest of the big companies, such as Amazon or Apple, who were not in charge of this competition of artificial intelligence, can clearly benefit and offer the same benefits as the rest to a portion of the cost.
rnrnnAct as an expert news reporters or journalists and create deeply engaging, well-researched, plagiarism-free news article BASED ONLY AND EXCLUSEVELY ON INFORMATION FROM THE ARTICLE BELOW, utilizing web search for relevant information and hyperlinking all external references directly to the contextual keywords within the blog body (NOT IN footnotes or a separate references section), including all provided quotes verbatim in quotation marks and attributing them naturally, seamlessly incorporating all multimedia elements from the original source, maintaining a sophisticated yet conversational tone with varied sentence lengths, integrating primary and secondary keywords organically, embedding relevant internal and external links, adding one table to summarize key points, strategically placing calls to action, fostering user engagement through fresh insights and meaningful analysis, and returning only the requested content without any additional commentary or text. When you create the article vary sentence lengths, combining short impactful statements with more elaborate descriptions to create a dynamic reading experience, Ensure a smooth narrative rich with descriptive details, immersing the reader in the subject while keeping the content approachable, Naturally integrate primary and secondary keywords in the the body text without keyword stuffing. Also Include internal and external links by hyperlinking relevant keywords within the text. All backlinks must be hyperlinked directly in the body of the blog, not in footnotes or a separate references section.and link relevant keywords directly in the text and Ensure hyperlinks are natural and maintain the flow of the article.
Do not place the sources at the end of the blog.YOU MUST HYPERLINK TO THE CONTEXTUAL WORD THROUGH OUT THE BLOG.
Include one table in the blog post to summarize key information or comparisons, helping break up the text and present data in a digestible format and vary Sentence Length: Mix short and long sentences to create a more natural flow and Be mindful of overusing certain terms or phrases, as this can signal AI authorship.
Do not place the sources at the end of the blog. YOU MUST HYPERLINK TO THE CONTEXTUAL WORD THROUGH OUT THE BLOG. Return only the content requested, without any additional comments or text.
The created article should be BASED ONLY AND EXCLUSEVELY ON INFORMATION FROM THE ARTICLE BELOW :nn:rnrn
Table of Contents
The artificial intelligence (AI) industry, once dominated by giants like Nvidia, Anthropic, and Google, is undergoing a seismic shift. In just a few days, the apparent advantages of these tech behemoths have crumbled. Today, fewer experts believe that training and using AI models require the resources of a technological giant.The monumental drop in Nvidia’s stock value is a stark illustration of this belief.
DeepSeek’s Chatbots: A Game-Changer
DeepSeek, a Chinese AI company, has emerged as a formidable player in the AI race. Their chatbots function similarly to other AI models, such as ChatGPT. Users simply sign up with an email and password, and they can start asking questions promptly. While DeepSeek’s chatbots may take longer to respond at times and avoid sensitive political topics like Tiananmen or Taiwan, they perform exceptionally well in Spanish. The best part? They’re free. DeepSeek’s business model focuses on charging only those who require professional use.
The R1 and V3 Models: A Leap Forward
DeepSeek’s latest models, R1 and V3, have caused a stir in the AI community.The R1 model, equivalent to OpenAI’s advanced O1, and the V3 model, comparable to ChatGPT-4, incorporate technical advancements that have disrupted the Western AI landscape. What’s remarkable is the cost efficiency. While training chatgpt-4 cost over $100 million,DeepSeek’s latest model was trained for less than $6 million.
Andrei Karpathy, co-founder of OpenAI and former Tesla AI director, described this budget as a “joke.” He added, “We must ensure that we are not wasteful with what we have, and this model seems like a good demonstration that there is much to review both in data and algorithms.” DeepSeek’s model is so efficient that it required only a tenth of the computing power of comparable models.
Cost Reduction Techniques
DeepSeek has mastered cost reduction in both training and usage. The company employs a technique called distillation, where the model learns from previous models. This approach, though ofen prohibited by service terms, is challenging to detect and avoid. By asking previous models questions and saving their answers, DeepSeek trains its models at a fraction of the cost.
Another innovative technique is the use of “experts.” Instead of activating all experts together, DeepSeek’s model decides which expert—be it medical, legal, or technical—to activate based on the user’s query. This targeted activation minimizes energy and computing waste, akin to connecting a small factory to a plug and a handful of generators rather than relying on a massive power plant.
The Impact on Users and Industry
These advancements have democratized AI.Users who previously paid around 20 euros for OpenAI models can now access nearly equivalent results for free. The expansion of AI is reaching its comforting stage faster than expected, where new technology becomes a common, accessible product.
All this innovation is happening in Hangzhou, a city near Shanghai. Despite the U.S. government’s export ban on advanced chips to curb Chinese competition, DeepSeek has proven that such restrictions are unnecessary. The company has embraced open-source principles, publishing detailed documentation online, making the need for advanced Western chips a thing of the past.
The Future of AI giants
This shift marks the end of the initial AI stage, where pioneers like Nvidia, Anthropic, and Google held significant advantages. Now, other major companies, such as Amazon and Apple, who were not at the forefront of AI competition, stand to benefit. They can now offer similar AI capabilities at a fraction of the cost, reshaping the industry landscape.
| Model | Training Cost | Computing Power | Performance |
|---|---|---|---|
| ChatGPT-4 | $100M | High | Advanced |
| DeepSeek V3 | $6M | Low | Comparable |
The AI revolution is far from over, but its trajectory has undeniably changed. DeepSeek’s breakthroughs have leveled the playing field,proving that innovation and efficiency can outpace sheer computing power. As the industry evolves, one thing is clear: the future of AI is more accessible and affordable than ever.