with tasks I can drag and drop tasks from one day to another edit them or add new ones it's like a full-blown app inside Claude I can copy its code or download it as a file that's amazing without an example we likely just get a plain list of tasks so if possible always provide
Поиск по транскрипциям
Google the latest AI headlines today and build a website around that right actually we need to use rode not that was client don't use clim use R code all right so it's going to start open up the browser there you go it's searching the latest AI news today inside Google it's actually...save on that and then we should be able to preview it now also the cool thing is because we're using Claude is the best for coding for front end everything else this website is nearly finished so this is the one about Julian Goldie which it built using the mCP servers from perplexity so the main thing here
right tools to get the job done. Let's teach our agents to steer and drive their own MacOss devices. What you're seeing right now is a cloud code agent operating a Mac OS device end to end. And I gave it a entire task list that I kicked off from a single just command. The great part about...everything we're going to cover. It's a lot simpler than it looks. We have our custom agents. I don't care if you're using Claude, Pi, Gemini, Codeex, Open Code, Cursor, CLI, whatever you want to use, unless you're building out your own custom agent harness. The interesting part really isn't your agent harness anymore
internal memos. We haven't seen the model yet. We haven't tested it. So take everything with a grain of salt. But the fact that OpenAI declared code red tells you something. They're serious. They're putting everything into this. And if they're willing to pause other projects, they must believe Garlic is the answer. So, what...right, let's wrap this up. Garlic is Open AAI's secret weapon. It's a smaller, faster, smarter model. It reportedly beats Gemini 3 and Claude Opus 4. 5 in coding and reasoning. and it could launch as soon as early 2026, maybe even sooner. This is the AI war heating up. Google won the last round. But OpenAI
matters. Most AI coding tools are either expensive or limited. Claude has usage caps. GitHub copilot by itself just does autocomplete, but Open Code connects GitHub Copilot to these premium models and unlocks their full potential. You get file editing, terminal commands, diagnostics, multi session workflows, and privacy controls all in one tool. And because it runs in your terminal
going to have to put maybe $5 or something on your Claude account. $5 is worth it if you want to learn how to code with AI. All right, so now Claude is working through the to-do list and it's going to create the backend server and hook up the API. It's going to take a while...know, it's tempting to go off and get a coffee while it's doing all this work, but there is a benefit to sitting around watching a code because you get to learn. It's still sharing these insights with me, right? And you also can see when it's actually going off track like using an older model
install everything separately the normal way to set this up is complicated and Technical but there's now an incredibly simple way to do it using Claude 37 sonit in Visual Studio code all you need to do is install Visual Studio code add the client extension and plug in your anthropic API key then
with MCP sometimes it doesn't work perfectly. So you've just got to be a bit careful there. Like for example, if you're using Claude and you can use claw code inside the terminal here. Like you can open up the terminal and just type in claude if you've already got it installed
scrolled through the website. I've had a look at what it's doing. resize the browser to check for mobile optimization and it's using that to vibe code out the rest of the page. All right. And then you can preview here. So it's actually just testing if it's optimized for mobile or not. And then...with MCP sometimes it doesn't work perfectly. So you've just got to be a bit careful there. Like for example, if you're using Claude and you can use claw code inside the terminal here. Like you can open up the terminal and just type in claude if you've already got it installed
focuses more on text and it does an excellent job at that. But I almost never recommend Claw to the average user who isn't into coding just because of how confusing and limiting their price structure is. So they basically have two plans. The first one is $17 a month if build annually, but that's $20 a month...these limits, even though I really don't use Claude all that often. So, unless you're a developer, a researcher, or doing a lot of coding, I really wouldn't recommend Claude. And if you are going to use it, you will probably need that max plan, which starts at about $100 a month. Next, let's look
going to chase coding does the rest of the model suffer I don't know I would argue that opus and claude those models are good at coding and they're good at overall writing so I mean I think someone said something that was really profound. Someone said the fact that Anthropic actually made their model to be somewhat...like this. Anthropic, their coding ability, if we look back at it, it is actually very good. We know that it is currently the best in the market right now. Software engineers say it is the best model to use. And most people don't realize this, but if you're actually using Claude to write articles, to write
also see that artificial analysis ran their own benchmark and we can see that they have the intelligence index. Now, this incorporates seven evaluations spanning reasoning, knowledge, math, and coding. And we can see that this model has jumped pretty much to the first in terms of the non-reasoning models which is like I said already a huge surprise...when it comes to reasoning models we can see that R1 is actually surpassing Claude sonet thinking in several areas but of course we know that Claude they are focused on just coding but nonetheless if this is V3 many people are speculating that V2 is going to be even more impressive and that the next thinking model could certainly
really does. It feels like Manis or Genpark or something like that. So it's now writing the to-do list to code that out. So this is approaching it completely differently from something like Claude. Claude will just go off and just create it in artifact straight away. Whereas with the Gentic mode inside Kimk...Right. So if we go to a new chat here inside Claude, plug that in. Do the same thing inside Kimmy. So we got them coding out side by side. It does look like Claude is way faster than Kimmy, but let's see. We'll see what you know which one gives the best outputs and it's also
gaining ground. In December, OpenAI declared a code red and pulled developers off other projects to focus on improving chat GPT. They needed to respond to Claude's recent improvements, particularly in coding, and it seems to be working. Elman told employees that chat GBPT is back to exceeding 10% monthly growth. More importantly, their coding product, Codex, apparently grew
unbelievable and so we're going to take the most complex well-built designs and documents an AI can build and put it into Claud Sonet which is going to make the best code we can build so paste this in so what we're going to do is we're going to use this prompt I just pasted...going to see what this pops out in a second okay looks like the code is completed uh generated a lot of code here which is awesome let's look at the preview in Claud son which is really cool if you click preview with any of your code you can actually see the code running in your browser
with any AI model that works with your coding tool. I like GLM 4. 6 because it's really good at coding tasks, but you could use Claude, you could use GPT, whatever works for you. Google also recommends using it with Gemini CLI. That's their command line tool. It's free and it's been fine-tuned...work really well with this MCP. So, if you want to use it for research tasks or general automation, Gemini CLI is probably your best bet. But for coding, I prefer Kilo with GLM. It just works better for building actual apps. Now, here's what I really love about this. It makes AI coding feel way more reliable
with any AI model that works with your coding tool. I like GLM 4. 6 because it's really good at coding tasks, but you could use Claude, you could use GPT, whatever works for you. Google also recommends using it with Gemini CLI. That's their command line tool. It's free and it's been fine-tuned...work really well with this MCP. So, if you want to use it for research tasks or general automation, Gemini CLI is probably your best bet. But for coding, I prefer Kilo with GLM. It just works better for building actual apps. Now, here's what I really love about this. It makes AI coding feel way more reliable because
through that as well in the meantime. So Claude me is basically a plug-in for claw code that gives it permanent memory across your coding sessions, right? So it's kind of like giving Claude a brain that never forgets. And the problem is like every time you close cord uh claw code, you open it again...manage. There we go. It's also quite easy to install because you can just get claw code set up in the first place. And then how it works, simple explanation. It automatically captures stuff, right? So every time Claude uses a tool reads a file or make a change, Claude memory captures it in the background. This also allows
understand that but I'm going to rate it number three Claude then deep seek then Gro then chat chti comes out last for coding chat chti is absolutely smashing it along with Claude for writing Claude is smashing it deep seek is not doing too badly version three and Gro as well these were drawing deep seek
give you Deepseek access for free. Deepseek just released version 3. 1. This thing beats GPT 3. 5 on coding tasks. It scores 66% on complex coding tests. Some tests show it beating Claude Opus. The interface looks like a browser with tabs. You can build multiple apps at the same time. Each tab is a different project. You just