Transcription
Hello. Today I will tell you whether you should switch to Open Code, which has recently gone viral on the internet and is presented as a killer of Claude-Code. Do you know what tires me out? Every day on YouTube I see videos with titles like "CLD Code is Insane." "Antigravity is Insane." "This agent can do anything." "LM Notebook is Insane." "Anything." "We have insy." The problem is that this doesn't help you choose the tool you will use. If you constantly switch between these tools, you won't be able to delve deeply into each of them. Today I will honestly tell you whether you should use Open C D. Well, I'll say right away that personally I use CLD-Code. I like it the most. Personally, I believe that at the current moment it handles complex tasks best. And the easiest way to use OPC almost limitlessly in the max subscription. For me, this is the easiest path, but for many of you, Open C D might be the best choice, which is why I decided to make this video. Recently, Open C D has made quite a lot of noise and has surpassed Claude-Code in the number of GitHub stars, 58,000 stars for Claude-Code and 77,000 stars for U-Code. At first glance, it seems, wow, some open-source project without a ton of money for marketing, unlike Anthropic, has overtaken Claude-Code in the number of GitHub stars. Amazing. But the problem is that the amount of noise around a tool doesn't actually tell us how good it is. Open C D is presented as a tool made by developers for developers, made with quality, with a higher quality UI compared to Claude-Code, and with fewer bugs. And additionally, this whole wave was fueled by a recent situation where Anthropic banned the use of their models via subscription through third-party tools. After this post, a ton of indignation began, and Open, in response, allowed the use of their models via subscription in Open C D. Who is making Open Code a serious competitor to CLDO D? People started writing that they are actively leaving CLD-Code, that Anthropic is not what it used to be, and by making their tool closed, they are losing many users and opportunities for expansion. And it's quite understandable why these people feel this way. This is indeed a serious response from Open Code and Open AI. Let's now see what Open Code looks like in practice and what its main killer feature is. So, I am on the official website of Open Cod D. Installation is as simple as possible. You just need to copy this command and paste it into your terminal. We are now in the interface of this tool. And using the Connect command, we can connect our LL provider. Now, pay attention, we can connect our GitHub Cilot subscription. We can connect our Chat GPT subscription. If you simply pay $20 for a Chat GPT subscription, you can use it directly in Open Code. You can also add your local models here, or models through Open Router, or in any other way. If your company prohibits the use of external models, you can add an LLM deployed within the company directly to Open C D. If you have any subscription, for example, to Antigravity, or you can simply use game authorization and use game for free, just like you can do it through gaming, you can do it through Open C D with all the advanced features. If you use Antigravity, then you can use Opus from your Antigravity subscription directly in Open C D. This repository details how to do this. Similarly, you can do this for authorization in your Google account and using gaming. Personally, I use Open Code as a second model, with which I check what D-Code has done. With Open Code, I check the plans that D-Code makes, and review code as a way to do this with another model. Open C D, just like Claude-Code, has sub-agents that can be configured. They are configured exactly like Claude-Code, just in a different folder. If you have configured agents for Claude-Code, you can take them and transfer them to Open C D without problems, adding details in the form of models and modes. And skills and Claude-Code are pulled automatically, and you don't even need to do anything. Thus, the functionality of Claude-Code is almost completely replicated in Open Code. But the main feature is that we can use absolutely any model we want. Let's try to play with Open Code and ask it to create a browser To-Do app. We send the prompt. I will use GPT 5.2. I would like to change the model to Codex. We edit the prompt. The UI of Open Code looks quite neat and beautiful. Some say it lacks the flourishes of Claude-Code. However, I have encountered them too, so I don't see any significant differences yet. It has also launched a questionnaire mode where we can choose our answer. Exactly the same as in Claude-Code. I will ask it to create extended functionality in React with storage in local storage and a design closer to To-Doist. Well, let's launch it and see what Open Code will do for us. With one prompt, we got this beauty. We can create tasks. I press Enter. Enter works. We have Inbox, we have Deep Focus. And these are all different projects. We can manage projects, create a project. I even have some labels. A task has a description, project, priority. Very similar to To-Doist. And we got this with one prompt. So, if you don't have access to Claude-Code, or if you realize that it is much more profitable for you to use some provider in Open Code without overpaying for Claude-Code, or if you have corporate restrictions that do not allow you to use Claude-Code. If open source is important to you, if you don't like the current UI in Claude-Code, if independence from a specific operator is important to you, then Open Code is your choice. If you generally don't care, you understand that OPСus is the strongest model right now, you understand that you already have a subscription. Or, if you don't care about all this and have no restrictions, then I highly recommend using CLD-Code. Personally, this is my choice. I don't have any fear of missing out. I also use Open Code, but now I understand that OPC is personally much easier for me to get in Claude-Code because I have a subscription to ClЛД, and OPCus, in turn, in my opinion, is currently the most powerful model for writing code on the market. The most important thing is not to fall into the trap of constantly comparing and testing tools, when you can't focus your attention on one tool and improve your skills in using that specific tool. Personally, I believe it's better to master one tool and do it deeply, without any constant paranoia and searching for other new tools. People often focus on the tool they choose, but, in my opinion, the methodology is primary. What prompts you write, how you manage context, how you use memory band. It's a whole art. This is what I talk about in my consultations. You can follow the link in the description and check out the landing page. If my videos have been useful to you, I have a lot of free material, you can follow the links in the description. But if you want to understand advanced AI usage without stress, as simply as possible, not just the features of Claude-Code, but how to use it and get results from it, then you can follow the link in the description and check out the offer. Let's summarize briefly and draw conclusions on when to use CLD-Code and when to use Open C. Personally, I use both of these tools because I have both a Chat GPT subscription and a CL subscription. However, I choose Claude-Code as my main tool because I work with it constantly and because it has a lot of Opus. But if from the perspective of choosing models, Open C D suits you better, you want to spend less money, if you have some corporate restrictions, or you like open-source projects, you don't like Claude-Code because it's buggy, then you can try Open C D - it's a really powerful alternative, and I don't think it's for nothing that it's become so hyped recently. Moreover, if you have a Chat GPT subscription, you can just take it and start using it for free. Therefore, as a second agent in synergy with Claude-Code, it works perfectly when I need to thoughtfully challenge something with another model. I use GPT 5.2 extra High and feed it what I need. There are also sub-agents, which is also very cool. So, here's my conclusion and opinion on this matter. The main thing is to remember that you don't need to constantly switch tools. You need to choose one, delve deeply into it, understand how it all works. And then, when you switch to a new tool, you will understand the underlying concepts. And it will be quite easy to do. It's like advising a programmer not to learn all languages at once, but to learn one language well, and then it will be easy for you to write in others. So, the next time you see that some tool is incredible, remember this video, think critically, make the right decision. Thank you very much for watching. You can subscribe to my Telegram channel, there are even more features there. Also, under the video in the description, I will leave my informative videos where I show concrete workflows on how to take and write code with Claude-Code, with Open Code. Everything I show there can be used in both tools. So, follow the links and watch. If I have provided you with great benefit, you can go and sign up for a consultation. Thank you very much for watching. Write quality code with neural networks. Yeah.