GLM 4.7 Flash + OpenCode + Antigravity is INSANE!

Key Takeaways

We have a brand new release of GLM 4.7 Flash today, which is a local coding AI agent and assistant. And we're going to be testing out today to see how it performs. All right. So, you can see here uh this like a comparable model to for example like GPOSS or Quen. And you've got GLM 4.7 Flash. Now, I've used GLM 4.7 already on the chat. There's a really good agent that's available for free at chat.z.ai using GLM4.7, but this one is different because it's faster and more affordable. In fact, you can actually run it for free. So, I'm going to show you exactly how to use this today and we'll test it out side by side. So, you can get access to this at hugging face as you can see right here. So you can for example go to the inference model and you can test this out. So if we say okay are you working we can test it out and type in a prompt right here and then it will come back to us using GLM 4.7 flash You can also see the benchmark evaluations. So in terms of the benchmarks here, let's compare it against Quen. So you can see here it's outperforming Quen by quite a long way. And it's [snorts] outperforming GPT OSS, which is the open-source model from Chat GPT, by quite a long way. Pretty much all of the benchmarks right here. So you can see here they've said setting a new standard for the 30B class GLM 4.7 balances high performance with efficiency blah blah blah and we're going to test it out. Um G says can you please link the toolkit you made a video about? Yes. So if you want all the links always inside the AI success lab which is a free community I put all the links. The reason I don't put the links on YouTube is just like it's easier to bunch everything inside here and then you can type in whatever you want. So for example, if you wanted to learn about GLM 4.5, you can just type that in the archive and then get all the links and the resources right here. So everything is inside the AI success lab link in the comments description. You can also see it's available on LM Studio as you can see right here. So let's test this out. And LM Studio is a way to host free local models like you can see comes in a nice UI. So if we download LM Studio and then we're going to install this like you can see. And then if we type in LM Studio, we should be able to get that set up. Just wait for that to install. And then you can see we can download this right here. So, we can download that. That is a huge file, by the way, like 16 gig. It might take a while to download that, but you can see how we can get access to that inside LM Studio. So, you can see 4 Flash 47 flash is available right here, and you can begin to use that inside LM Studio for free. Also, if you get the pre-release of Olama, this new update, as you can see, you actually get access to GLM 4.7 Flash, as you can see right here. So, if you want to get access to this, so let's try and install this. absolutely ages. So what we can do is we can download like the latest version of Lama and then from there we can install GLM 4.7 Flash. But the problem is you can see here it's like it might take an hour to download that. Let's see. It's getting faster. In the meantime, let's see what questions we got. Oh, this just nearly done. Let's go. All right, here we go. So, let's check that one. 14.3. There we go. So, you can see it downloaded now. So, what you have to do if you want to get it running on Olama, you go to version 0.14.3 of Lama. Download the zip if you're on Mac. Then what you're going to do is run that once you've downloaded it. Make sure you're on 0.143. And then inside the terminal, you just run this command as you can see here. So, alarm run GLM4.7 flash. And then you can get access to this model directly right there as well. It is a huge model so it's going to take a little while to download. That's essentially how you can get access. There is GLM 4.7 inside open code as well but I can't see 4.7 flash. Maybe we have to upgra up update it. Let's see. Can't see it now. Let's have a look on the comments what we got here. Why is bro dancing? Um, walking. Why? To get the steps in. Stay in shape. Why not? What are we doing right now? Can you run a quantized version of GLM 4.7 Flash? Yeah. So, you can run you can run it on a Lama and LM Studio like we were talking about. By the way, 4.6 Flash is trash. I've not tried it out. So, let's see. Team Insights, please stop doing AI videos as soon as I see you published. Honestly, like if you look at the views and if you look across every single channel, if you look, you know, whether it's on Facebook, Reddit, if you look on Twitter, if you look on YouTube, the videos that tend to perform the best are avatar videos, right? And typically my human videos perform worse. And so, in general, people vote with their views and their clicks. And, you know, AI avatar videos tend to outperform the human version of me, sadly. What else we got here? How can you type whilst walking? Very nice setup. Thank you very much. I'm on a treadmill. Yes, getting the steps in. Getting the steps in. And we got a new member as well. Welcome, Chlo. Thank you very much for joining. So you can see that new model is loading right here. It's taking a little while. There we go. So, we got GLM flash now running on LM Studio. So, if we go down here, we'll select the model. We'll try GLM flash. It says model loading was stopped due to insufficient resources. Continuing to load the model would likely overload your system. So, I actually can't run it locally, sadly. Um, just so you're aware, I'm on a Mac with a M4 chip. So, I'm on a Mac Mini, Apple M4 chip. I can't run it locally, which is kind of annoying, but better than crashing during a live stream. And you can see here the 4bit runs fast on a base M5 32 GB laptop. So I think if you have a powerful laptop, this is going to work nicely. But if you're just using a normal one, um, this is going to be pretty hard to use. See, only place you can really start using it then directly is inside hugging face. You can use it online. They've got an inference provider right here. So you can just test it out inside the chat there, see how it performs, etc. But for local deployment, that's going to be pretty tough unless you've got a super powerful laptop from what I can see. I might be wrong, but that's just what I've seen so far. Let's see how other people are using it. Oh, there we go. So, we can actually use it on open router as well. So if you don't want to use it locally or if you don't have the power to use it locally then you can just load it on the API inside open router and then if you want to start coding with it just load up your favorite ID like this. Then we can see if we can use it inside client. Let's have a look. Obviously, if you don't run it locally and you're running it with open router, there's going to be a cost to that, but it is super cheap. Like look at that. 0.07 uh per million input tokens. 200,000 context. Not bad. And then if you wanted to run it locally, you can select O Lama and then select the model and then go from there. Right? If we want to use it inside open router, I think we can use it like this. So let's just try this out. Yeah. So you can select GLM 4.7 flash like this and then you just get an open router API key. So let's find that And then once you've got your API key, just use it inside your ID. So you can plug it in right here in the settings of client. Hit done. And now we have GLM 4.7 flash running. Right. So if we say okay create just to test this out, we'll see how it is. So we'll say like create an SEO calculator. Let's see how it performs and also how much it costs to build out the project. You see it's a reasoning model. So it's using thinking mode 200k token context. A little bit slow but maybe that's because it's just been released. It should be fast to be fair. And also if we have a look inside terminal now we have GLM 4.7 flash running with Olama. I do think that's going to be very slow on my laptop if I try and ask it something but we'll try this out. So, you can see I've typed in the prompt right there for a llama. And it's it's not it doesn't seem to be working on the API, which is weird. So, it's not working on terminal. It's not working on the API. [laughter] Good start. It worked on hugging face, but it was quite slow to respond. Ashkit says, "Love your content." Thank you very much, sir. Odoro says, "Just arrived. What are we building? We're waiting for GLM 4.7 to work." It is very, very slow, though. Look at that. I mean, it's cheap, but this is nowhere near the speed of of what I would expect for a flash model. And Chlo says, "Good content. Love what you're building. Thanks very much for joining. I'm glad you're in the boardroom, too. Appreciate you being there." Um, I actually have to shut down because it was just being so slow. So, I'm going to shut these tabs down. Yeah, this is way too slow on anti-gravity. Look at that. It's lagging my whole computer. So, overall, not an ideal model to use, I would say locally, if you don't have a powerful laptop. Let's go to the chat here and just see how it performs. I think next time I upgrade or next time I get a, you know, a new laptop or whatever, I'm going to have to get some crazy RAM on it. So, let's try this out. We're going to try Opus versus Flash. As you can see right here, I'm going to shut down O Lama. Shut down Studio. Let's try and try and get it working again. So now we have GLM 4.7 Flash working here and Claude Opus over here. And it looks like, let's have a look here. Looks like it totally failed on me. So, GLM4. This is It's funny when you see this because like I think most YouTubers would never show you this, but you can see we've tried it on pretty much everything. Lama, LM Studio, ant to gravity with the API key for open router and open router inside the chat itself and it's just not working like anywhere I've tried it. So maybe it's just been released and it's like, you know, still slow or everyone's rinsing it, but so far I would I would never use this model. Like you're going to get way better results if you just go to chat.zai, Z AI select GLM 4.7 and then build something out right there. Right? So, for example, let's say you want to build um a website. You can see some of the stuff we built previously using GLM 4.7, not the Flash version. And look at that. Like super nice. Uh the UI looks really nice. The website looks nice. We literally just gave it a prompt to create a beautiful landing page based on uh some information that we plugged in and it created a much nicer page. So, I would still say for 99% of people watching, unless you have a super powerful laptop or you're really really insistent on running local models, you just you wouldn't use GM 4.7 Flash. Doesn't make sense. So, at least I've been transparent and shown you what works and what doesn't. If you do want to see my training on how to use GLM 4.7, you can get that inside the AI success lab completely free link in the comments description. So, you can see we have loads of trainings on like how to use this for AI SEO and also how to build stuff with this, you know, all our best resources and step-by-step tutorials. So, I would recommend starting there if you want to learn this stuff. This is completely free community that connects you with 47,000 people as you can see. And then if you haven't already, check out the AI profit boarding link in the comments description. This is basically an amazing community full of serious AI builders who are focused on growing and learning and scaling with AI automation together. We actually have a daily accountability group like you can see where you can check in, post your goals, avoid information overload, avoid shiny object syndrome, and uh avoid overwhelm and just focus on your goals for the day. You can also post inside the community, get help, get support whenever you need it. And you can see it's not just me who's winning with this stuff, right? So loads of people are just getting results with AI automation and they share their progress inside the win section of the AI profit boardroom. Top of that, you get four weekly coaching calls per week. So you can jump on these coaching calls live, get help, get support whenever you need it. And then also you'll get my best trainings inside the classroom. So, for example, you can see the six week AI automation masterass that shows you how to go from complete beginner to expert with AI, plus how to build your first AI agent in under 5 minutes. On top of that, you'll get my best playbooks for automations on X, Twitter, AI avatar videos, and also you'll learn how to get more clients for your agency using the agency course. Additionally, you'll learn how to rank number one inside AI search engines and how to grow a YouTube channel based on what's working for me right now. So, feel free to get that link in the comments description to the AI profit boardroom. Stark DevOps says, "Have you tried Cerebrass?" I've not tried that. Let's try it out now. Let's see. Has he got GLM 4.7 on there? Let's try this out. So, it's got GLM 4.7, but it doesn't seem to have flash on there, which is what we're looking for. So, I think even Cerebrass, you don't seem to be able to get access to GLM 4.7 flash, which is what we're looking to test out today. All right. So, today we're going to test out this new skills update for anti-gravity. So, Google Anti-gravity is an IDE where you can code locally for free and you can basically download it for free. You can get access to browser use agents. You can even, for example, fill up PowerPoints and spreadsheets for you. It's a really powerful tool for vibe coding. I've built some amazing stuff with it already. But the cool thing is recently they updated with a skills update where you can basically create these skills MD files that are like agents that show exactly how to um do different things, right? And specialize in different things. So, for example, like autonomous and aentic stuff. You've got stuff for creative and design AI agents and LLMs, etc., right? So, what we're going to do is we're going to install this directly on anti-gravity. So, I'm going to show you exactly how to do that. We'll see how this performs. Now, also, what's pretty cool about this is that it even has access to different stuff. So, you can for example install documentation for notebook and things like that. Right? So if you have a look down the list, you'll see that it comes with hubspot integration for CRM and also we have marketing ideas right here and additionally we have the MCP builder for building MCPS and also notebooks. It's going to be cool to test this out. So let's try this. We're going to go to anti-gravity here and I'm going to start up a new project, new window. So, let's open up a new folder. And then from here we're going to go to the terminal and install these skills. Now you might think this looks technical if you've never used terminal before, if you're not a coder, etc. I'm not a coder at all, but what you can do here is you just go to terminal, new terminal, and then you paste in the commands from the GitHub. Right? So, you can see the GitHub right here. If you want to get access to all my best trainings, including the links from today, go to the AI success lab link in the comments description. Then go to January, go to my notes on anti-gravity. We've actually got a full training tutorial on this, plus a bunch of other stuff you can use. And I'm going to in insert the resources from the GitHub. So here you can see. And literally all you do is you just copy and paste this into a new terminal. There we go. All right. All right. So, if we go back to anti gravity now, we've got those agents installed. So, if we go inside here, you can see we've got all the skills set up like this. That was super easy. And then what we can do on the chat on the right hand side is we can tag these in So, for example, here we've got this front end design skill that we can test out. So, let's try that out. First, when you go to antigraphy here, we'll type in app front end design and tag in the agent skills. And then from here we're going to say okay build out a beautiful landing page for my book link building mastery right insert an image of the book and also some information about it which we can get from my website and you can use your existing assets to train the AI on what you want to build and how you want to build it. So we'll copy that plug this in. So you see the image is now inserted here and then we're going to grab the information from this landing page. As you can see, we'll go back to anti gravity. Hit enter. And now we should be able to start creating this landing page. So let's see what we get back using the new skill. Now what would be interesting as well is that we take the exact same prompt and we test build this out without the skill. Right? So, if we open up a new window here, we'll call this no skills, and we'll compare the before and after and see which one develops the best All right. So, that's finished creating the front end. Let's see what we got on the page. Pretty nice. Look at that. And by the way, let me show you the old version. Right. So, if you go back to julianoldi.com Yeah. Basically, the goal with these skills is to create better outputs, right? So, we'll see. But hopefully, we've got a better version of the page using the skill versus no skill. We'll we'll test this out in a sec. So this is the page with the skill and then this is the page without the skill. This one right here. And you can see it's still editing the page right here. So both of them are still generating. And also if we look inside the skill MD files you can see how they're written right here. So for example if we go back to facel here you've got a full skill for facel deployment which means that you can actually deploy directly to facel using this skill which is pretty useful. You've also got for example the front end design skill here too and the front end developer skill. And then you can see, okay, here's when to use the skill, here's a new component checklist, etc. And there we go. That's our landing page built. As you can see right here, definitely looks a lot more modern than the original. So, this is the original page that we have currently live on our website. And it's not bad. Not bad at all. But if you have a look at the newer version using the skill for front end, it just looks a lot nicer, a lot cleaner, a lot more modern, interesting. I'd probably switch them over to be fair. Like this looks much better. As you can see, um there's even like these nice animations. It's quite clear what to do. It's a nice little squeeze page. Gemma says, "So the skills are imported into anti-gravity." That's it. So you just use the terminal command to import the skills. And then when it comes to actually hosting the project, I think it would host everything. So it would host all the files for the actual project which is down here. So you can see got the index, HTML, JS, and CSS. And then on the left hand side, the whole folder of the project, anti-gravity skills build has the skills, the agents, the scripts, everything else already pre-installed. You can also see here that these aren't just for anti-gravity. So you can also use them inside cursor or claude. You got skills for Gemini, etc. So it's kind of like something you can import on whatever CLI you prefer. >> [clears throat] >> Now, it's also got the skill for notebook. I'm not sure what it can do. Let's have a look here. So it says use this skill to query Google notebook LM notebooks directly from core code and it's a research assistant skill. So you can interact with Google notebook to query documentation and if I mention notebook LMS explicitly it should trigger this skill. Let's see how it goes. So, I'm going to say, can you use notebook LA to research Julian Gold's link building mastery book? Let's see if that works. So he's actually using that skill. Setting up the environment to use it. You just need to authenticate. It's going to say, "All right, let's authenticate." Plus, I don't have a specific notebook URL just yet. Let's see what we get back. So, it's going to ask me to sign in here. And we just got to grab my password. and then it should be logged in I think. All right. So it says great the authentication was successful. Notebookm library is ready to go. Now we just need to provide a URL. to query notebookm. Let's try this. Let's see if this works. So, we'll create a new notebook. Pull in some sources quickly. Then we'll copy the link to it and give it to anti-gravity. See, so you can see here it's actually operating notebook. It's asking questions etc. you can connect directly with it and then it's like right I've analyzed notebookm added it to our research library it's a comprehensive resource covering SEO strategies link building tactics etc so it's done the research on the book using notebookm and then it's like right what would you like me to extract from the research so you can connect anti-gravity you directly to Notebook LM if you want to use this as a research tool, which is pretty cool. I've never seen that before inside uh anti-gravity. There's also this Loki mode which is quite interesting. So Loki is like a autonomous system I think. Let's have a look. So, start up in a box, sub agent, driven dev, dispatching parallel agents, planning with files, skill creator and developer. Let's try that. So say create a landing page about link builder mastery using the research use loy mode and made the design beautiful. Then it's using notebookm to help us. Plus it's analyzed the skill MD files for loi mode and everything else. Now it's looking at this continuity MD file as well which is interesting. So it's done the research with notebookm then it's going to design the architecture then it's going to implement it do the visual styling and the verification right so it's basically got like a a status list and checklist right here is right here's the next steps Now it should open it up. There we go. And look at that. Nice little landing page created. So it's got my name at the top, the information. It's even found that image. It's created a personalized landing page with the Goldie method talking about all the content from the book. So for example like uh Harrow link building podcast outreach etc. Landing page looks nice too. Your videos have me addicted to collab coding with uh anti-gravity 100. It's so good. Anti-gravity. The more I use it, the better I get at using it as well. And there's just so much you can do with it, right? It is pretty amazing for this stuff. If you're having issues with using the skills, just let me know where you're stuck specifically. Is it importing the skills? Can you see all the skills on the left hand side? If so, it's just a case of tagging them in. Right? So, you can put um at and then whatever skill you want to use, right? So, for example, let's say I want to use notebook. Well, I just use at@ notebook and then tag in the relevant skill like that. So, that's how to set them up and get them working. AntiGravity is now just testing out the page on mobile to see if it's good or not. DevOps says it's just a matter of time before Google becomes the main API provider. Yeah, I would agree with that. I think like, you know, it's it's only getting better. It's only becoming more powerful. And also, they're just giving it away for free, right? It's crazy. Uh I don't know any other tool that's as powerful for vibe coding as anti-gravity. And it's free, right? It doesn't cost me anything to use this. Plus, it records the video of the agent using the page. Like you see, it's all looking good and it works smoothly. I can show it on a live stream. No issues. Jenny says, "Can I just ask Antiggraft to install those skills?" So, you would insert them inside the terminal, but if you're not 100% sure, you can always look through the files online before you install anything and just do your own diligence, do your own research, that sort of thing. For me, I'm happy being like the guinea pig to try it out and then if you see anything you like, you can integrate it into your own workflows. And you can see here we've got Loki mode running. So, this is enhancing it like you can see with the notes right here to tag the skills. So all you do is you put the at sign here and you type in the skill that you want. Right? So you can look on the left hand side here and be like right okay well I want to test the AB test setup right so I put at AB and then we can see the skills or for example at AB test and we've got AB test planner right there and you can see the directory of the skill as well right so agent/skills app store optimization bioden says Claude slays. Yeah, I would agree. But you can use Claude directly inside the API right there. So if you want to use Claude Sonic with this, you can use it there. Or you can just go to terminal and just run claw code inside the terminal. If you type in claude here, there you go. Do the skills have to be inside the workspace or globally? So you can install them globally or you can install them locally. It's up to you how you want to do it. Ah, see apparently as well you can just go inside anti-gravity here instead of the terminal and then ask it to install things for you. I haven't tried that actually, but yeah, I should try that next time. Thanks for the tip. So, let's have a look at this landing page. This is the finished product. As you can see right here, it's looking really nice. Taken the custom image. It's got some nice animations. It's written all the copy, designed it, and it's also used the same brand colors, right? So, it's taken the brand colors from the book and then just added that inside the page so it all fits nicely together. Added some icons right here as well. And then it's got the CTA to to book in a call, right? Along with a couple of testimonials it's taken from notebook and then plug that right in here. Nice. So that was super easy and simple. And then if you want to deploy it, you could use netlifi or you could use the skill for for sale, right? So you could type in app for sale here at for sale deployment and then just say okay deploy this. What that will do is give you a sharable custom domain or subdomain that you can share with people right here using the skill MD file. Right. Usually I use a netify mcp but if we can use cell why not and then I think we need to get an access token for Netify. Sorry for forel. The other option you have is you can just say okay deploy this with Netlify. It will call on the MCP tool for Netlify as you can see right here. And then boom, we've got that ready to go. Let's click open on that. There we go. Right. So, we vibe coded the whole page out using anti-gravity notebook. We use the skills for Loki as well as the front end development skills. And then you finally have the page right here. So that's basically how to use the skills from anti-gravity. Like you see all the video notes from today are inside the free AI success lab along with the resources and terminal commands for installing this. This is a completely free community that connects you with 47,000 members as you can see. And inside the classroom, if you go to January, you'll see the full guide on Google Antigravity, how it works, the resources, and the terminal commands. Plus, we got a full step-by-step guide on exactly how to use Google Antigravity as well. So, thanks for watch Thanks very much for watching. And if you want to get access to a community of serious AI builders who are focused on growing and sa saving time and scaling with AI, you can get that inside the AI profit boardroom link in the comments description or go to apiprofitboardroom.com. We have a daily accountability group where you can stay laser focused, avoid information overwhelm, and avoid information overload using the daily accountability group where you post your goals and then focus on whatever you need to achieve. You've got the win section here where you can share your wins, share what's working for you. You can see all these people absolutely crushing it. So, for example, Chris built his first custom GPT. Jay absolutely smashed it, winning the weekly challenge. You've got Steven crushing it with Miniax, etc. And also inside the classroom here, you get access to all my best trainings. Now, if you ever get stuck and you want to jump on a live call, you can see the live coaching calls right here. So, you can jump on these calls. And then also inside the classroom, you can see that we have a full road map on exactly how to go from beginner to expert with AI automation for your business. Plus, you'll learn how to build your first AI agent in under 5 minutes. As well as that, you'll get all my best playbooks for Twitter, for newsletters, for shorts, for AI avatar videos. And then also we have a full course on how to get more clients, a road map on how to rank number one with AI SEO and also a full six week map on exactly how to grow and scale your YouTube channel based on what's working for me. So you can get that all inside the AI profit boardroom. Plus, you can personally DM me. I answer questions inside there every day as well. So if you ever get stuck, you can post a question and then we're always happy to help and guide you and and respond to you as you can see right here. So feel free to give that link in the comments description to the AI profit board. So, let's see what questions we've got so far. So, Damian says, "I'm struggling with the Netlefi MCP in anti-gravity. Whereabouts he stuck? Just let me know." As you can see right here, all you need to do is install it, connect your personal access token, and then call upon it inside the chat whenever you need it. I have this really cool SOP skill for the AI before building apps. Yeah, share it in the boardroom. Would love to see that. Great information, brother. Got to run to finish that website workflow I was talking about in the boardroom. 90% done. When you're free, check the message I left about Claude. Awesome. See you inside there. See you inside the aircraft boardroom. I've not even learned how to use MCPS yet. MCPS surprisingly simple actually. We have a full guide inside the boardroom on how to use them and set them up and that sort of thing. As you can see right here, we actually have a full training on MCPS and that sort of thing. It's a full 1hour training. You can use the archive inside the air profit boardroom to get access to this stuff. Right. So you can see all the community posts about this stuff. And then also inside the classroom here, you have all these different trainings with video tutorials on how to set up MTP agents. Great content as always. Happy to help. Sam says, "I'm new here. Is this stream available to rewatch?" Yep, you can view it inside the live section of YouTube on my channel. Go to the boardroom. It's gold, y'all. says uh yeah thank you very much and then can you share me use case how we use anti-gravity plus open code I'm still doing the SEO training in what yeah there's some good SEO training inside there too can you share the use case on how to use anti-gravity and open code let's see one Yeah, let's get that set up. All right. >> [clears throat] >> All right. So, let's talk about how to use open code directly inside Google Antigraity. Now, Google Antiggravity is free, very powerful tool. You can see a page that we just coded out right here. So if we open this up, you can see we just built this landing page literally in a few minutes using anti-gravity. And the cool thing about anti-gravity now is like it can code, it can vibe code, it can use MCPS to deploy websites live. It's a very powerful IDE. It's free. You know, I never seem to hit the limits with Gemini Flash. And you can also get access to Claude Opus and that sort of thing. Plus, you get all these skills as you can see on the left hand side for pretty much everything you can think of even for example like using PDFs or powerpoints and that sort of thing. Right now, how do you use open code? So, what you're going to do is you're going to go to extensions over here and you can see that you can get the extension for open code like this. Right? So, what you want to do, make sure you install that. Then once you've installed it, you'll see something like this. Right? So just to be clear here, you go to extensions. Then you make sure you have Open Code installed like this. And then you can go and start a new session over here. Right. There's also a few different variations of this. So for example, you've got open code GUI as well. And also I think you can install it inside the actual terminal as well. Let's have a look. And then what you can do is as you can see right here, we've set up open code and it's using that to delegate tasks, Right. So for example here we've said use open code to improve the CSS and now it's editing the CSS for this particular project using open code. Right? So it's got this open code run command right there. And you can see it's beginning to code out inside the CSS to improve it. Now if we for example type in the terminal here we can also use open code. So if you just type in open code run like you can see over here and then you give it the command. So, I don't know. Open code run 5 + 5. It will answer you using the AI inside the terminal right here as well. So, there's two different ways. I would honestly say like the easiest way is just use it inside the chat here. But if you want to use it with the terminal directly, you can over here. You might say, why would you use that? I mean, honestly, it's mostly because you get access to more API credits, right, for free, right? So open code is free and then if you ever run out of limits with open code uh sorry with anti-gravity you can use open code directly inside the terminal to keep coding out right to to bypass API token limits. In fact, the number one comment I usually get on YouTube channels uh on YouTube videos when I'm coding with anti-gravity is that it has too few tokens to use, right? And so if you're using open code then you bypass that and you can keep coding even if you run out of credits on anti-gravity. Let's see what questions we got here. Are you all confused a bit when combo anti-gravity and open code something it feels similar or redundancy? So again, you just use it if you're running out of credits, right? Or if you want to access different APIs. What have we got going on? We're using open code directly with anti-gravity. Plus, we've shown how to install skills as well and use this as an agent inside anti-gravity. And Alex says, apparently Google AI Studio starting to AB test again. Rumor said a Gemini 3 Pro checkpoint. That'll be exciting. It's been a while since a good update came out from Google, so I'd be excited to see that. So, we've got the new updates right here, as you can see. So, it's verifying the CSS improvements. The agent is running in the background to UI test it inside a separate Chrome. It's inspecting the design elements. >> [clears throat] >> And if we go inside here, you can see this is the new page with the new CSS for this particular project. So just to recap here, what have we done? Well, we previously had a project that we built out using antigraphy. We ran open code inside the chat but also inside the terminal and showed you how to use it. We showed you how to get more credits on whatever you're coding by using open code instead. And then also I've shown you how to for example update and improve the front end using a combination of open code and also anti-gravity. Now the cool thing is if you use this inside the chat directly, you could also use skills with open code, right? So you could tag in for example an MCP server and use open code with that directly or what you could do is you could get the AI agent to test out the changes from open code. Right? So if you look inside the Chrome browser here, the AI agent is now looking at the front end changes the open code is made. Like for example, this hovering feature over here. So you kind of get the power of open code with anti-gravity directly and then you you never run out of tokens essentially because you're using a combination. You can also, for example, use client and other Visual Studio Code extensions inside the extension section here. Install them and then also bypass token limits using that too. But that's basically how it works. So, I'm just going to stop the agent in the background right now. Let's stop that. Force close. This is the updated page. The improvements are subtle, but it looks nice with the hover animation and everything else. And that was pretty easy to set up. Plus, this is all free, right? Anti-gravity is free. The MCP servers are free. The skills that you can import from GitHub are free. And then also, Open Code itself is free. So, it's a really powerful combination. And you can also deploy this to subdomains like we've seen um earlier in the video. So, thanks so much for watching. If you want to get all my best training on anti-gravity, you can get that inside the AI success lab. link in the comments description. It connects you with 47,000 people who are interested in coding and building with AI. If you go to the classroom here and then you go to January 2026, you can see a full video tutorial on anti-gravity, the full resources from today, how to set this up, etc. We also have loads of training inside here if you search in the archive for open code. So, you can see all these video tutorials if you want to learn how to use open code and how to build with it and what other cool stuff it can do. as you can see. And then if you haven't already, check out the AI profit boardroom. This is an amazing community of people serious about growing, scaling, and building their business with AI. Right? So if you want to learn, for example, how to use open code or for example, you want to get live coaching for you or your team, you can do that inside the calendar right here. Inside the community, you can ask questions whenever you get stuck with something and it's a very active community, so you can get help and support whenever you need it, right? So, for example, we actually have this daily accountability group where you can write down your goals, avoid shiny object syndrome, avoid information overload, avoid overwhelm, and just post your goals and be like, "Right, I'm going to focus on these today." And we actually check back in and make sure that you actually implement them, which is pretty cool. Also, inside the win section, you can see all this cool stuff that we're doing. People sharing their wins, their journey, what they're creating, etc. Lots of people getting awesome results as much as me with AI automation. As you can see, Dylan is sharing his progress and got some awesome wins with NAM, for example. So, if you want to learn and grow with AI, this is the best place to do it. And also, we have a full automation course that shows you how to go from complete beginner to expert with AI automation, plus how to build your first AI agent in under five minutes. On top of that, you'll get my best trainings, for example, for creating AI avatar videos, screen shares with AI clones. Like you can see we've got training on how I automate X. Bear in mind like yesterday we reached 457,000 people in one day with X. So it's a really powerful tool if you know how to use it. And then also you can see here that we've got a full agency course on how to get more clients, how to scale your agency. You can watch the coaching calls and the recordings there. If you want to learn AI SEO automation, we have a full section right here and also show you how to build a full YouTube channel from scratch showing you what's working for me right now. So feel free to get that link in the comments description to the AI profit boardroom or go to aiprofitboardroom.com.