Speaker 1: StudyIQ IAS, now ready, affordable. Let me tell you that China has shocked the whole world. Yes, friends, here you can see on the internet, on social media, there is a lot of discussion about DeepSeek. You can see the news, How Small Chinese AI Startup DeepSeek Shocked Silicon Valley. No one would have thought that a small Chinese AI company will do such a big miracle. It means that the chat GPT, Google's Gemini, can also beat them. This is what has been shown. In this video, we will know about DeepSeek, what is DeepSeek R1, why is it being discussed, what are the most important things in it, I will tell you all that. Let's move ahead, but before that, let me tell you that the Republic Day Sale is valid till 27th January. If you take any batch on StudyIQ related to UPSC, whether it is our pledge batch or for 2026, Prelims, Mains, Interviews, everything is prepared in it. And if you just crack the prelims, then the entire fee is refunded. To join, use this code ANKITLIVE so that you can get maximum discount. And the link is given in the comment section. And if you want to attempt MCQs, then go to my Instagram page. You will be able to attempt all the MCQs in the stories there. So let's start. First of all, let's see what exactly is happening. See, two years ago, you will remember, OpenAI's chat GPT came to the market. And the kind of performance it was giving, people were surprised that what is happening. In this, whatever we want to do with chat GPT, it does it. And after that, a lot of improvements have been made. So because of this, it seemed that China is behind in the matter of AI. That is, America will move ahead faster than this. Because first, Microsoft and OpenAI's chat GPT came out here. Because he invested in Microsoft. Then besides this, Google's Gemini came. And after that, Meta, Amazon, many American companies They were launching their AI model. But what is happening here is that the Chinese players They also tried, like China's Alibaba, Baidu They invested billions so that they can also get behind it And at the same time, they can generate their own AI model But now what is that things have turned Suddenly, if you look from social media to everywhere There is a lot of discussion about China's company So, the first question is, what is DeepSeek R1? DeepSeek R1 is a state-of-the-art reasoning model developed by Chinese AI startup DeepSeek. It is a simple thing, like when we say chat GPT, chat GPT is an AI model, an AI tool. You go there, whatever you want, you have to write an application, you have to do anything in today's date. You simply type, you get the answer. Then there is a lot of talk about OpenAI's O1. So these are all AI tools that you can use. And who made it? It was made by OpenAI. Similarly, the startup in China called DeepSeek They launched this AI model tool called DeepSeek R1. And the kind of performance this AI tool has shown The kind of exceptional performance shown on competitive pricing due to which it is being discussed a lot and their main purpose that is their aim is that the rest of the AI players in the market whether it is OpenAI's O1 or ChatGPT can beat them in a way can compete with them here DeepSeek R1 is an advanced language model which is constructed by hybrid architecture Also, it is performing better than its predecessor. Because let me tell you, this Chinese startup, AI startup DeepSeek They had launched another AI tool earlier called V3 But it was not that successful But now they have launched R1 It is becoming very successful And it is being told that the large-scale reinforcement learning, chain of thought reasoning You will get to see it very precisely Means whatever you want, you will get the output as you want. And let me tell you that there are two versions in it. One is DeepSeek R1 and the other is DeepSeek R10. And it is being told that this R10 has such a remarkable ability that again it is performing better than all the other AI tools in the market. But the question here is where is the origin of DeepSeek? Because suddenly we are getting to know about DeepSeek in the market. Look, there is a discussion about DeepSeek R1, but very few people know about this company. Actually, its headquarter is in Hangzhou, China, and this company was established in July 2023. And who established it? Liang Feng, he has studied from Xinjiang University, and he is basically an information and electronic engineer. Also, it is being said that OpenAI's Sam Altman's goal was to develop an Artificial General Intelligence which is bigger than the human level of proficiency, which is better than the human level of thinking, which has precision. So, the same aim is here of Liang. He also says that we want to make such an AI model. But the question is, how is DeepSeek gaining so much popularity in such a short period of time? Because it was launched in the beginning of this month, on R1. So, the most important feature of DeepSeek is its cost-effectiveness. Let me tell you how. For example, if you see OpenAI's O1, it charges $15 per million input tokens. Means $15 per million input token $15 means in rupee terms, it is around Rs.1250 And the open AI for per million output tokens charges $60 If we talk about DeepSeek R1, its price is very less It charges only $0.55 Means it is around Rs.1250 and it is hardly around Rs.50 And for the output, they charge only $2.19 So, from this you can understand how cost effective it is Some people might be thinking that this is cheap, so isn't it giving good performance? It's not like that The people who are using this, they are saying that it is cost effective But it is giving the same performance as OpenAI's chat GPT, Gemini or OpenAI's O1 And in some cases, it is also being said that its performance is better than OpenAI's O1. That is, 97% success rate can be seen in it. Along with this, DeepSeek has also launched 6 compact versions of R1 model. So that it can run properly on a laptop, mobile, and in different places. And in fact, let me tell you, another important aspect of DeepSeek is that they developed this model in just 2 months. Can you imagine, the big companies of the world have been working for years to develop it. China's small startup company develops such a model in two months that Google, Microsoft, OpenAI can beat everyone. And in how much money? They have invested only 6 million dollars. If we talk about OpenAI, Google, Microsoft, they are investing billions of dollars here. So you must be understanding from this that why it is becoming so popular. And because of this, Satya Nadella, CEO of Microsoft has remarked and said that I think we should take the development out of China very very seriously. Many people should take it seriously, that's what he has to say. Anyway, if we talk about user experience, people are posting their experience on social media. For example, there is a user on Twitter named ZeroEdge. He checked the same thing on both the places. As you can see on Twitter, on one side is chatGPT's O1 and on other side is DeepSeek's O1 In this, input was given to implement a rotating triangle with red ball in it So if you see chatGPT's O1, red ball is out of the triangle. means the result is not that perfect whereas in DeepSeek, you can see the red ball in the rotating triangle So, it is clear that in many ways, it is beating other AI models and AI tools Now, the next question here is that this AI model DeepSeek R1 has been launched Is it an open source model? Can people use it further? See, it has been released as an open weight means the researchers will be able to study it and will be able to further build on its algorithm. Yes, you can't call it fully open source because the training data has not yet been released in the market by them. But let me tell you, because they are a Chinese company, slowly China is also trying to open itself, trying to bring a little transparency so that it can show the world that it is not behind this. For example, if you look at Alibaba, they have published more than 100 open source AI models in recent months, which supports 29 languages and you can use it for coding, mathematics, etc. means you can build further things The report published states that China wants to put itself ahead in AI development on a global scale Like last year it was said that 1328 large language models were present all over the world Out of which, China accounted for 36%. That is, 36% of the world's 1328 large language models was owned by China. So, in a way, China is the second largest player in AI development behind the US. And yes, you can say that it may be a little behind the US for now. But what DeepSeek has done, because of that, it can now go beyond the US. That is, it is growing very fast. And China's influence in AI is increasing more and more. So somewhere this question arises that should India not move forward in this? Should it work? What do you think about this? Tell me. But before going, let me tell you that if you want to use DeepSeek R1, like you use ChatGPT, in the same way, you have to go to chat.deepseek.com, you have to verify the email there, enter the address and click on DeepThink. Then you will be able to use it. So this is it friends, I hope you have understood why there is so much talk about DeepSeek And before going there is an interesting question for all of you Can you tell me, you have given three statements related to AI, which statement is correct? You all know the right answer to this, you will get it on my Instagram If you are not there, no problem, you can take the right answer from Telegram I keep putting all the MCQs in the stories here If you have a PDF, then come to Telegram You will be able to connect on different social media And as I told you, if you are preparing for UPSC, especially during Republic Day Sale You can take the best batch with extra discount for your preparation. And use this code ANKITLIVE I hope you liked the video. See you next time, till then thank you very much. See you soon. And friends, before the end of the video, As you all know, UPSC is the most prestigious exam in the country. Where more than 10 lakh aspirants fill the form. But the selection is only of 900. So the competition is tough, but your dream is also big. P2I Pratigya 2026 batch, where your journey will be easy and with this Republic Days special 55% discount, you will get the whole course in just 27,999. In this, you will get the top educators of India, Anil Swaroop's Mentorship, 18 books study kit and guaranteed prelims clear, full fees refund and cash reward of 11,000 rupees to the top scorer in the test every month. What you have to do is go to the studyIQ website, click on UPSC Pratigya badge and once you click on buy now, you can see discount codes. There you have to type Ankit Live and you will get the lowest price. Batch will start from 27th January.