Deepseek, the Chinese AI that has destroyed Chatgpt, a victim of a large -scale 'cyber attack

Deepseek, the Chinese AI that has destroyed Chatgpt, a victim of a large -scale ‘cyber attack

Deepseekthe company behind China that in a few days has revolutionized the technological sector, is being object of a cyber attackas explained on its support page: ‘due to Malicious large -scale attacks Against Depseek’s services, we are temporarily limiting records to guarantee the continuity of the service. Existing users can log in as always. Thanks for your understanding and support, ‘says Deepseek.

The platform had already had difficulties throughout the day, but it has been in the last hours when they have increased. Deepseek’s AI, who launched its new language model R1 A week ago, it is assuming a break in the technological sector, which has been overturned with artificial intelligence. In the United States, The app reached the first position of the most downloaded in the App Store, ahead of chatgptand has also been the cause of important losses in Wall Streetbeing Nvidia The most affected, after being vanished 500,000 million dollars of stock market capitalization in just one day. This has been the greatest loss of stock value in a day in history. The curiosity for the app has shot.

In this context, users have not been surprised that the platform had difficulties due to the excessive number of users. The company has published that it began to investigate what was happening with performance this morning, Spanish time. Three hours later, he warned of cyber attack. You have not given more explanations, although The most likely option is that of an attack of denial of service.

The reasons behind Depseek’s success are Offer an experience similar to that of the most advanced language models in Chatgpt, which are paying, but for free. It is also a software from open source that anyone can audit and use for their purposes, downloading it and executing it locally, in addition to the option to use the app or the web. But, above all, It is very cheap.

Faced with the multimillion -dollar investments that are constantly talking Restricted access to the most advanced hardware of NVIDIA or AMDdeveloped its previous model V3which is still available on the platform, for just 5.5 million dollarswhile OpenAI He had to invest 100 million in GPT-4. According to the Chinese company, the development of R1 has cost between 90 % and 95 % less than its competitors, compared to those who have stood out in various performance tests for these artificial intelligence models.

And he has done it with a more limited hardware than that they use companies such as Microsoft, Google and OpenAi, but which seems to have taken more advantage. Deepseek claims to have used only 2,000 graphics cards of nvidia to train v3, while the main market models use from 16,000 up. That is to say, Deepseek has achieved the level of Silicon Valley companies, but without requiring the same more advanced investments or hardwarewhich explains the impact that Nvidia has received, whose evolution will be key in the next few days.