Speaker 1: Hello everyone, welcome back to Vamsi Bhavani If you are following the AI news, AI technologies, and the different updates in AI closely, there is information about Deep Seek AI from the last 4-5 days If you are not following the news or not understanding the context properly Vamsi Bhavani is here So, what is Deep Seek AI in this video? Why is Deep Seek AI suddenly becoming so famous? Why do you see a comparison with ChatGBT? And finally, how can you use Deep Seek AI? Let's discuss everything So, stay tuned and watch the video till the end First things first, Deep Seek AI We have multiple LLMs like Demony, ChatGBT, Cloud Similarly, this is an LLM-based AI tool You can use this as a chat platform or chat interface Or you can use it via their APIs So, it is similar to ChatGBT, right? The main thing for us I mean, why do we hear the name ChatGBT compared to this? Around this, I mean, there is a lot of political commentary also There are a lot of stocks in the US, up and down, right? I won't go into that politics, right? In this video, I will focus on giving you the context of Deep Seek So, first of all, if you open the website DeepSeek.com In this, the main thing If you see, in the heading Deep Seek R1 is now live and open source It is good up to there Rivaling OpenAI's model O1 So, they are saying that we are a competitor to OpenAI's O1 model Rivalry, I mean, see Rivalry is not a fact here So, what they are trying to say is We are saying that we are a competitor to that model, right? So, to understand how they are a competitor First, let's understand what OpenAI's O1 model does So, just go to OpenAI's O1 page, right? OpenAI.com slash O1 So, recently, OpenAI has released this O1 So, what is this O1? In some tasks, we need more reasoning and thinking power Even in the regular day-to-day works, right? For some tasks, let's say, we have to start the bike We don't need logic or reasoning for that, right? Since we have a lot of time, we'll just start it Let's go, right? But let's say, you want to solve a maths question Or you want to solve a competitive programming coding question In that, we need to apply a lot of logical and reasoning These are the observation skills that we need to apply, right? So, in the same way, when you want to use AI to perform a task If you need more reasoning in that These reasoning models will help a lot And OpenAI O1 is one such reasoning model If you see, we have developed a new series of AI models Designed to spend more time thinking before they respond So, simple, right? Directly, not just based on a recent context or a limited set of context If you spend more time on reasoning and thinking We call such models as reasoning models And in those reasoning models, one is OpenAI's O1 The other is DeepSeq So, DeepSeq is that kind of a ChatGPT competitor Now, why a competitor, right? First things first, they are already beating GPT-4 in their multiple metrics Okay? So, do we have only these two in our reasoning models? No, we have a lot Then, why is DeepSeq all of a sudden? Like, you know, compared to OpenAI's ChatGPT Now, I think, in the Apple App Store DeepSeq is trending Like, it's number one in positions Crossing ChatGPT Why? The reason is, already DeepSeq ChatGPT, that is, OpenAI's models Has crossed in multiple metrics That's number one Number two, if you want to use OpenAI's O1 You can just try it using ChatGPT Plus You can use the Plus subscription And if you want to use it unlimitedly You have to pay $200 monthly Can you imagine? You can use OpenAI's O1 for just $200 Huh? 1.5 crores That's a lot, right? But, the same DeepSeq users Are providing it completely free That's the biggest game-changer When we are doing business, He's in the middle So, this way, wherever money is involved If you are finding something cheaper At the same time, if the performance is good What will you choose? Obviously, the cheaper and better performance model That's why DeepSeq is trending so much Right? And, why is this political commentary? Because, DeepSeq is a Chinese company Chinese platform Right? So, if you want, let's talk about the political commentary But, I don't really think It can help you or me a lot We just want to use AI tools And, get benefit from it Whoever does it Right? So, how to use this? Right? Let's go into that point First of all, click on start now So, we have chat gpt or chat interface That kind of interface is available So, due to large-scale malicious attacks On DeepSeq services Registration may be busy I told you, around this Let's not talk about that But, a lot is happening, guys Because, I mean, things are getting really tricky here I mean, AI in this competition US, China, this competition Cool So, let's try Because, I'm not sure if If we try with a new account Will it work or not I already did it with my personal account But, let's try with my Vamsi Bhavani contact account Let's see if registration works Let's try that Hmm So, we are in So, simply Similar to chat gpt Log in using your gmail Let's try to ask a question Right? Let's ask a coding question involving logical reasoning Let's assume there is a hole If you solve that hole with dynamic programming How will the approach be? If you solve it with backtracking How will the approach be? Let's try to understand that Right? So, here you If you click on the send option directly You will get the answer But, that reasoning model will not be used So, for that reasoning model to be used You have to select DeepThink R1 here Right? So, if you hover, you will see that Something is coming here Let me hover it again Use DeepThink R1 to solve reasoning problems Right? For example, I want a poem I want a song I want to understand a song I don't need R1 for that I need R1 only if reasoning is involved Since, reasoning is involved here Let's, you know Select R1 And click on send Here, the biggest plus point is This model explains its thought process So, the answer you see right now Again, I am not editing anything So, basically it's the same time It will take this much time to think actually So, if you want, you can minimize this This is not the answer It is just the thought process On how this model thinks that question So, once the thinking process is over Which is still continuing So, if you see If a person, an experienced programmer Asks to solve this question The way he thinks The whole thought process is done by this model Right? That is the power of reasoning models Since, they are giving that for free This is trending So, if we want to use this further I will make videos But this video, I want to keep it simple for Everybody who wants to understand DeepThink So, yeah Thinking is done So, after thinking for around 47 seconds The answer was given Anyways, right? So, wherever If you think and take the time to answer It will be more optimized That too, in logical questions Optimization in terms of reasoning Answers will be more accurate Right? So, let's minimize this thought process Let's see the answer So, what are the core concepts? He gave the background So, conclusion Backtracking is preferred solution Offering a balance of simplicity and efficiency Dynamic provides DP provides No practical advantage Here, due to problem structure Right? So, I mean, that's the correct answer too But you If you are a programmer You can also learn from this thought process Right? That's how, you know, this is So, they gave us the entire interface Like a chat jeep So, it's easy to use Try it You will understand Like, just What you have to do is These questions Where reasoning is required In those cases Select this once Try it without selecting You will see the differences And additionally What is the capability that we don't have in O1? Reasoning model Plus search Has not been enabled yet They will do it It's not difficult Because search is already there But in this Already enabled Let's say For example Let's say a live cricket match is going on Let's see if there is any live cricket match Going at the moment For example Let's say it's going on Right? So, basically What you can do with this search Okay Looking at this live match performance Because it will refer to all the links in the top search Right? By referring that Which team do you think will win? Why? Provide your analysis By saying Reasoning R1 Plus If you enable search and ask It will use the current data So, it will be relevant And because it uses reasoning The answer will be more accurate Did you understand? So, involving search and using R1 Will be even more helpful Not available in O1 at the moment But Shouldn't be an issue for them I feel Because there is search There is reasoning They will add The only big challenge is Budget Will they cut down their cost by looking at this or not? Let's see So, overall I think you got a good idea about Deepsea If you got it Hit that like Share it with your friends No one in Telugu can do this kind of information Please share And If you are new to the channel Please do subscribe Thank you very much Jai Hind Jai Bharat