DeepSeek R1 лучше ChatGPT. Сравнение, тесты и мнение Claude AI. Объективный тест нейросетей.

DeepSeek R1 лучше ChatGPT. Сравнение, тесты и мнение Claude AI. Объективный тест нейросетей.

Show Video

let's compare the Chinese outside of DeepSeek and ChatGPT, but it won't be you and me who will evaluate them, but Claude will evaluate them, let's go, look, I will conduct an experiment models to use in ChatGPT there will be 1 model in DeepSeek we will have a model a R1 aa dfink that is, a model with the ability logical reasoning means that Claude will evaluate, that is, we will now ask ChatGPT and depcic questions and the answers we will receive accordingly, we will import into Claude and ask the question deck to ask which of the neural networks coped with the task better, I think that this will be more objective because if we carry out the analysis ourselves, we can give preference to someone else and not that non-neural network which, in the opinion of, for example, Claudeac, will cope with the task better , then look at all the products that I will use in my requests I will take from my telegram channel the telegram channel where we have collected more than 10,000 different products for Chatje 5 either for DeepSeek or for Claude for any artificial intelligence plus there is a huge knowledge base , a video base where we are where you can just watch and fully understand any neural network in order to understand how to use neural networks for your business, for your personal life, for your freelancing, depending on what you do, that is, we have a huge database knowledge database database of pronts database of tables database chat for communication where you can interact with another participant and accordingly got to learn a little more about the neural network the link to the Telegram channel is under this video be sure to go register and it will be there all the information about how to get into the database , how to gain access to this database , then look at the 1st Promt which we will use it Promt to create a content plan, a content plan tailored specifically for rils for rils or well, Tiktok Instagram, it doesn’t matter, you can upload these videos for the mode to shorts without problems, that is, we send this request to ChatGPT4 so I will immediately check that I do not have any user instructions enabled , yes I do not have any user instructions connected so I just send this request aa to ChatGPT aa o 1 and I immediately send the same request directly to DeepSeek version R1 without the search function and send this request to depsic here we can see how ChatGPT reasons, how depcic reasons by the way DeepSeekk reasons was able to reason several days ago in Russian he was already arguing in English, apparently more people are interested in English, well, look gpd chat has already created almost the same table for us with the databases dep sigget thought at first and is now just starting to create the Claude table accordingly , we are still waiting for him, we will use it now , we will now upload these 2 tables here, respectively, I will probably make them in the form of screenshots, that is I’ll create screenshots with these tables and then upload them to Claude aa, but first I’ll set a contribution, that is, look, the jeepty chat has already dealt with it, let’s not waste time let's not read aa and DeepSeek also, in principle, coped with the task we now have these tables these answers we will send directly the contribution 2 of these considerations 2 of these screenshots they are now being loaded and in Claude we set the following Promt these are 2 tables of content plan for creating video content Instagram or Tiktok created using the DeepSeek path charge, do analyzes for each of the tables, tell us what the advantages are , what are the disadvantages of each of the content plans, make it a list by tell each of Zizi in the finale which neural network coped with the task better in your opinion and for some reason we ask Claude to conduct such an analysis of the content plan and mmm and tell about the advantages of the disadvantage of each of the content plans and highlight in his opinion which content plan is better which neural network accordingly coped with the task better then look at ChatGPTadvantage: very detailed structure with a large number of categories , which allows you to cover many content formats good and arhat and Ir hah irhichical organization of information with a clear division by content type a detailed description of each format with specific examples includes a wide range of content from educational to entertainment disadvantages may seem too complex overloaded for beginners although we never told him that we are creating a content plan for beginners maybe we are creating for professionals the table structure is somewhat confusing with a large number of nested ones categories there are no clear time frames for the implementation of the content plan there are no specific success metrics for each type of content this, by the way, is very important and DeepSeek advantages a simpler understandable structure with a clear division into the type of specific content specific practical recommendations for the implementation of each item includes time guidelines , for example, in 30 seconds in 15 minutes, contains additional tips for optimizing content, for example, using CTI subtitles, and providing ready-made examples of hashtags, specific tools, less coverage of the format compared to charge drinking, some points could be disclosed in more detail, lack of information about the frequency of publication, in my opinion, let’s conclude, let’s move on, in my opinion, dipsix will cope with the task better for the following reasons, the plan is more practically oriented, easier to implement, the structure is more understandable and logically includes specific time frames and metrics provides additional practical advice on optimization recommendations despite less information the plan seems more feasible and less overloaded although the GBT chat provided a more extensive plan the DeepSeek approach is more practical and convenient for real-life application especially for those who are just starting to work with video content that is, he tells us that the content plan created with the help of DeepSeek is more focused on beginners, it is simpler, it is more understandable and, accordingly, it will be easier for us to implement content plans from ChatGPT, according to DeepSeek it is overloaded it is more complex it includes some so stop look at the networks hey aa contains additional tips for optimizing content for example City hey yes aa did he understand the content plan from deepsico and the content plan from ChatGPT because deepseek is actually very similar to ChatGPT and um and Claude could just stupidly get confused in the interface and say that, for example, this is the content of the Deepsico plan because Claude doesn’t have access to the Internet and he doesn’t know what kind of Internet he has at the moment ChatGPTinterface and this accordingly, he might have thought that it was from Dipsico, so let’s ask him this, although he should notice the logo, I asked him, did you understand correctly that the content plan from Shaji 5 is a screenshot of such and such, that is, I told him the name of this file and the 2nd screenshot is a content plan from DeepSeek and now look what it says to me, sorry for the confusion, I see my mistake, in fact, both images show 1 and the same service names in the browser header and ChatGPT and DeepSeek but judging by its real addresses to the interface, these are different parts of 1 and the same plan content are simply shown in different tab windows 1st image chat jeep ChatGPT com shows a longer expanded table of plan content 2nd image DeepSeek chat dot com shows a more compact version of a similar one information, I apologize for the incorrect information, could you clarify which of these screenshots was created using the potikva charger using DeepSeek, this will help me carry out a correct analysis of the data, look, yes, he got confused in our request , but let us remind him again and, for example, say that I sent him this screenshot again, said that this is a screenshot from Charge 5, which means thanks for the clarification, now I see that the 1st image is really from charge 5 and 2nd depsig let's conduct a correct analysis of ChatGPT advantage more structured approach advantages depsic that is, he again describes the advantage to us and again says that DeepSeek coped with the task better because he suggested more practical plans that can be implemented, the information provided is more accessible, that is, he did not abandon his original answer yes G DeepSeek in this in this short competition in Claude's opinion he wins let's move on to the next task the next thing I want to ask ChatGPT and DeepSeek is to conduct an experiment for me, that is, to conduct a little research for me, imagine that you are an expert in marketing on social networks, in particular Instagram, tell me 30 disappointments 30 wishes 30 fears that aa experiences here I now need to register my target audience, even just basicly, yes, as a target audience I simply wrote that this is an entrepreneur, a blogger, experts promoting on social networks , in particular Instagram, imagine this in a tabular format where the x-axis is from 1 to 30 and the axis igrek this is a disappointment desire the fears of our target audience we send it all to ChatGPT and send it all to deepsik the same thing look he owes me aa create a little marketing research we we can see how ChatGPT thinks we can see how DeepSeek thinks, respectively, the task is a little more complicated , but oh I think that they will both cope with this task now, so look ChatGPT started to perform the task DeepSeek also began to perform the task , then look, they both begin to create a table, so something like that DeepSeek, of course, has a slightly strange table ChatGPT has already started the table with the table and started working, which means it’s an interesting approach, of course deepsik broke the table, that is, he did everything in more than one table, he aa broke the table into parts chanji 5 live look does everything in 1 table and aa disappointment desires and fears of our target audience small number of likes and comments with great effort desire to have high involvement community activity fear of remaining unnoticed in the huge flow of content DeepSeek look yes here it is for some reason I divided this table into 3 parts, I hope of course that now he will take it all and combine it no he created 3 separate tables for each section although I asked him to create it in tabular format but he didn’t say that it should be 1 table, okay, the jeepetidja chat created exactly 1 table, so let’s send it all back to Claude and see who coped with the task better, I entered the following request into Claude, I’ll now send you 2 screenshots in turn created using the chat jeep Dee and Dipsy, do an analysis of each screenshot and tell what the advantages and disadvantages of each of the neurocity are and sign that these are screenshots response from Chang Ji 5 , respectively, he began to analyze the advantages disadvantages, I’m waiting for the 2nd screenshot, give it to him right away Let's send the 2nd screenshot, this is the answer to the next one, don't grow deepsik, I sent him in the form of 2 such screenshots, a more modern structure of the answer the information is divided into 3 separate tables, which improves readability and navigation according to ChatGPT, let’s also ask him in the final which neural network coped with the task better in your opinion , and although he, look here, has already written such a small conclusion that the jeepty chat provided a more emotionally charged and an interconnected answer while dipsix emphasized the technical and business components of the chadzhipti better showed the psychological aspects of depcic and the practical problems in the opinion solutions, both approaches have their value and the choice between them may depend on the specific needs of the user, but to the final question which of them corrected the task better, he did not answer, but said that at least Charge 5 is probably stronger here in his opinion, yes, look in my opinion, chatpetiz coped with the task better for several key reasons, the value of the approach , the depth of analysis, the practical value of usability and versatility, this does not mean that depsic’s answer is bad, that is, ha ha, we are all tolerant neural networks and say that this is not it means that your answer is bad, it means that someone else simply coped with the task a little better than you, he is simply more specialized and technically oriented, however, to solve the task of creating a comprehensive overview of the disappointments of the desires of the fears of entrepreneurs, the chadzhepitiya approach turned out to be more effective and practically applicable, that is, in this task, in the opinion Claude's ChatGPT wins , and the 3rd task, I think it shouldn't be associated with tables, but it's still the same work with text, which means look at what elements of the text structure are most important for increasing conversion in online sales write the ideal structure of a selling post for social networks, let's sign it, what in particular for Telegram how to learn how to teach artificial intelligence to reproduce this structure in generated texts and let's send this request to DeepSeek and send the same request to ChatGPT and compare their answers accordingly, that is, if To put it in simple language, what we are asking here is chargipti and deptik, we will ask them to create a structure for us for a selling text so that in the future we can create , for example, using this structure your selling texts in Telegram on Instagram, it doesn’t matter, you can also ask for a structure, for example, headings , or ask him to create his headings based on some texts or headings for our advertising company, this can also be done , but let’s ask him to do this for now and then we’ll see if we will have time so as not to delay this rils, we will also ask ChatGPT for it and ChatGPT you see there is some kind of error, the DeepSeek service is overloaded oh well , that’s where it begins let's send our problems to DeepSeek again I have you I got an error now and it’s very strange look the error came out in ChatGPT and the error came out in dpsic at the same time that is, it’s very interesting why Changi 5 and DeepSeek simultaneously gave an error, that is, ChatGPT said that it can’t answer the question right now, there was some kind of error, either it’s a problem with my Internet or they are somehow interconnected, this also shouldn’t be ignored then look, uh, DeepSeek begins to describe the structure of ChatGPT for us, he also described the structure for us, plus ChatGPTaa not only described the structure, I answered the 3rd question: how to teach an artificial cart reproduce this structure uh and gave us some results deepsik also said that how to teach and generate such texts gave us an example of a pront quite simple pront and accordingly also gave us an answer , let's now send both answers to Claude and put it I don’t know, maybe the final point uh, who coped with the task better, but once again I was tensed by the fact that uh, both of them hung at a certain moment until I send the next prom, I’ll send it to you now 2 screenshots one by one on the screenshots are the answers from Charge 5 DeepSeek at the request of Promt and I also gave him the request Promt that I gave to Charge 5 in DeepSeek aa and do an analysis of each screenshot and tell what are the advantages and disadvantages of each neural network answer, make it a list for each of them and then look at how ChatGPT will analyze it, let's look at the disadvantages of some redundant information in the 1st section which can overload the reader the lack of visual formatting, for example, highlight indents for a better perception of the structure, well, it’s very strange, yes, although all the indents are there, and the answers from Charjipit from DeepSeek are a less detailed description of the learning process and, compared to Charjipit, the lack of a detailed explanation of each element, the structure is not disclosed, alternative models comparing the answers of both neural network chat jeep ti gives a more academic structured approach with a deep study of each element depsic offers a more practical approach with modern elements and with specific examples, charge jeepty better explains the learning process and depsic better takes into account the specifics of Telegram as a platform, both answers complement each other can be used in combination to achieve maximum results , but we will still ask him which neural network coped with the task in your opinion better and why in my opinion DeepSeek handled the creation of the task better, here are the key reasons, the answer is more focused on real-life applications; specific examples were used in Telegram , and take into account the specifics of the platform; the modern approach of the structure takes into account current trends in online sales information is provided in a more compact and structured teaching methodology, a more practical approach to learning is proposed through the analysis of real posts, academic elements are included about testing for Valdats results, an emphasis on specificity and a ban on water in the texts, although chajik type provided a more detailed academic answer for practical application in the context of selling posts suitable for Telegram DeepSeek turned out to be more effective, applicable in real conditions, but something like this guys, look at our tests, well, it’s probably quite simple , a template test for fairly simple industrial processes that I, for example, use very often in my life, in order to create selling texts, create a content plan, create ideas for content, accordingly, according to all these parameters, Claude said that the free depsic is much, well, not much, but it copes with the task no worse and somewhere even better, that is, in 2 tests he said that the task from depsic coped with the task better and he accordingly wins the paid chat gpc what will the guys from Open night do lower the subscription price add some new functions about this we think that We'll find out soon, I hope that this video was useful to you, be sure to give me a like under this video Igor Zuevich was with you subscribe to the channel subscribe to the Telegram channel all promts can be used for both ChatGPT so for DeepSeek so for Claude for now

2025-02-04 23:43

Show Video

Other news

SMU vs. Virginia Tech Full Game Replay | 2024 ACC Men's Basketball 2025-02-08 22:29
The unseen world of 3d printing at 1000fps! 2025-02-08 06:39
Primitive Technology: Flywheel blower smelt/Monsoon begins 2025-02-07 20:07