next release is also from Claude. Super interesting one. If you're watching this channel, you probably already know about Claude codes. what many people consider to be the best AI coding assistant and many people also use it for marketing tasks product management chat GPT alternative but it has been notoriously challenging to use especially for nontechnical people because
Поиск по транскрипциям
different worktrees, we can see that Gemini is done over here. Codex code is still running because code is a fork of Codex, which is worth remembering. And Claude Code is still running over here. Okay, so now I think the agents are done over here because it's saying comparing agent results and it's running this built...better API/UX decisions. And I think that's true because Claude usually makes better UX decisions than Codex CLI does. Or I can do option C which is basically rerun them all. And I think the reason it gave this particular approach or this particular options is that code is actually still running for some reason. I guess that
compare for example the benchmarks of Claude 4, Opus 4 and 4 versus Gemini 2. 5 Pro. You can see the benchmarks here. So for agentic coding, Claude 4 is absolutely smashing. Gemini 2. 5 Pro 63% from Gemini 2. 5 Pro versus Claude 4 at 79 and 80%. If we look at aentic terminal coding, Gemini...going to plug this into Gemini. Use canvas. And then we'll also do the same inside Claude 4 as well. But I'm going to switch to Opus 4 because that's the most powerful model for coding. We'll plug these in and see what we get back. All right. So if we go side by side over
thinking. Flip it on before you hit enter. Once it's on, you immediately notice the difference. Instead of spitting out a final answer, Claude starts narrating its thought process. It might sketch out pseudo code, run calculations in an artifact, or list out options one by one before picking the best one. If you give it a data...spot. That kind of transparency saves you time and confusion. So when should you use it? Anytime accuracy really matters or the task is genuinely complex. Debugging code extended thinking helps Claude methodically test possible fixes instead of guessing. Analyzing data goes through each angle before concluding. Writing a nuance essay weighs arguments carefully. On the other hand
going to get into some additional science. So this isn't only like anecdotal stories, I promise. So before AI, my job was think about a problem, write code, test it, ship it. I was the creator, the maker. That's what drew most of us to engineering in the first place, the act of building. For me, the analog...know this treadmill probably better than almost anybody on the planet. He says, "Take a breath and try to keep up with it just the last few months. Claude Code ship sub agents, then skills, then an agent SDK, then Claude Co-work. OpenAI launches Codeex CLI, then GPT 5. 3 Codeex, a model that literally helped code itself
testing this side by side versus root code. And Rukode is like Manis, very similar to Manis. They both use Claude, right? They both use Claude 3. 7, but inside Root Code, you can actually have this use your browser directly and it can use your login. It can write code and build and automate things locally. Plus
data on me and it understands me so well, right? So it's one of the best things for that. However, if you want to code something out, then I would recommend Gemini or Claude. It's very rare that I actually use Chat GPT for coding out. I actually tried to use it for an NA10 workflow this morning...this, And that's a bit easier to do than something like this. And obviously as well, you can get access to Claude inside Claude. ai as well, but it's not really a coding agent. This is more just like better for writing, better for learning. and it can do basic code but it's not going to code
seems to be a lot more impressive what we can actually do is we can copy these artifacts so once you actually build a piece of code or a tool inside Claud you can publish it and then you can grab the link for artifacts now if for compare side by side this is the page from 3. 7 Sonic...preview it and then self iterate on that kind of like a coding agent which is super good in the meantime whil that's coding out let me show you some other stuff so Claude 3. 7 Sonic
comparisons right here in terms of context window I would say Gemini is winning for coding Gemini is winning if you're trying to build something cheap with a big context window but the actual outputs inside Claude 4 when we compared content when we compared apps etc. Claude 4 is winning and also for a gentic autonomous use because...compare, for example, the benchmarks of Claude 4, Opus 4 and 4 versus Gemini 2. 5 Pro. You can see the benchmarks here. So for Agentic Coding, Claude 4 is absolutely smashing Gemini 2. 5 Pro 63% from Gemini 2. 5 Pro versus Claude 4 at 79 and 80%. If we look at aentic terminal coding, Gemini
agency you can roll this into your services. Uh just like a few days ago I was telling everyone how teaching companies how to understand claw code and set up their own workspaces as a service was a thing. This seems to be like the new frontier on that. So yeah, it's just it's so great...Yeah, it's already not even the end of 2026 and it feels like it's been a year. So especially on the education side, you could have taught Claude code alone. Then a week later, you could teach cloud code and co-work. And now code, co-work, and cloudbot. And then you could probably keep stacking
near as good as Gemini and Claude again. So, you know, if I had to basically switch between these, I would say that Claude and Gemini are kind of like neck when it comes to coding. chat between Grock nowhere near the same. And if you saw my video the other week about Grock 4. 1 as soon...Chat GPT working over here and we've got Claude Opus over here. Now what one thing you're going to see is that Claude has already finished coding the tool before anything else. Right? So look at this. We've got Claude over here. It's already created the news. We can click fetch news headlines using Opus
compare 2. 5 Pro preview which was previously my favorite model against Claude 4. Now, Claude have just released two new models, Opus 4 and Sonet 4. Sonnet 4 is pretty good for everyday tasks. Opus 4 is the best for coding and you can see the benchmarks here...compare, for example, the benchmarks of Claude 4, Opus 4 and 4 versus Gemini 2. 5 Pro, you can see the benchmarks here. So for agentic coding, Claude 4 is absolutely smashing. Gemini 2. 5 Pro 63% from Gemini 2. 5 Pro versus Claude 4 at 79 and 80%. If we look at Agentic Terminal coding, Gemini
them, and then how do I do outreach? Okay, and your outreach should follow this system. Then you're going to deliver the work with AI. So, Claude, long form copy, strategy, coding, uh, chat, GBT, codeex, coding, really good. Gemini research brainstorming the creative suite they have um Sora and V3 video generation inn is for like back
they got quite eager and started implementing the solution which I just deleted to make sure it's fair uh for the next step. Uh whereas Gemini CLI and claude code did not start implementing the solution partly because cloud code is in plan mode over here which prevents it from being able to write any code. Anyways
session and there's a ton of context in there. When you get a lot of context into Claude code and starts filling up your window, it actually can make Claude hallucinate a lot more than it does before. I hear a lot of people on X complain right now, oh, Claude's getting worse. No, that's a skill
this a lot. So, let's see. We're going to open up a new window here. Open up Claude, right? So, you need to make sure you have Claude code installed. Got loads of training on that. And then we're going to type in this prompt here. See if it actually works. Dr. says, "Given the unique reasoning
hasn't even created the CSS yet, whereas Claw Code with fast mode is already fully deployed the website on a subdomain using skill boss. That is insanely quick. Just got to be careful with tokens, like I said. But if we go back to our token usage on Claude AI here, refresh the page. Bear in mind, previous...What I've actually done is inside the AR profit boardroom, if you go to claude fast over here, you got the full link to claude codes guide on it, how to use it, etc. There. And then also, just
something else. Carlos as well, he's another member of the AI profit board. Shout out to Carlos. Says, "If we already have Claude code downloaded on our app in our laptop in VS Code, do we have to download everything again? " Not as far as I'm aware. So, I think it would just automatically update or you should
find a few more prompts for starter projects like this that you could try yourself right away. So go and try out Cloud Code today. And if you enjoy it, sign up for the Claude Max subscription, which allows you to build many applications like this without you having to worry about...cost every time you add a new feature. That's everything I have for today. In case you want to see me build out the larger project with Claude codes, you can check out this video. And I hope you have a wonderful
процессов или обработки конфиденциальных данных, имеющих большое значение. Кроме того, они внедряют кэширование с мгновенным откликом для повышения эффективности, а их SDK позволяют создавать собственные агенты для кодирования. Теперь Claude Code напрямую интегрируется с популярными инструментами разработки, такими как VS Code и Jet Brains. Иными словами, они предпринимают скучные, но крайне важные шаги, чтобы убедиться, что эта технология не только