Devstral: FREE API + FREE MCPs is WILD!

Today we're going to be looking at Devstral, which is a free API you can use to code and pretty much build anything you want. So for example, I actually built this and this in the space of like literally a few minutes using Devstral completely for free and the stuff actually works. As you can see, it's really easy to build out. So let me show you exactly how to do it.

You can also, for example, connect it to stuff like Firecrawl MCP, which allows you to basically connect to the web to any agent using firecrawl mcps and a combination of devstral. So let me explain exactly how to use this, how to get set up, etc. And also you can host this stuff locally and I'll come on to that in a minute as well. All right. Now, this is basically like a very small model that is super cheap but really fast. And if you want to start using it, you can go into something like Visual Studio Code here and then just go to your settings, select open router, and then inside open router, the way that you're going to use this is you are going to select Mistral and then Devstral, right? And there's a free version here. So, you don't even have to pay for it. It's completely free, right? So you can select mistraout devstr as

your API. And then for example, you can see we built out this AI growth calculator. It didn't cost us any money. We literally just said build this out. And the cool thing about this is as well if you're building something for example like the tools I showed you earlier, then it will step by step plan that out with you directly and show you exactly how to build it. So let's take some examples here. I've actually got a list of 100 ideas inside the AI success lab if you want some ideas on what you can build with this and that sort of thing. So let's say for example we want to create a random password generator. All right, we're going to go inside Visual Studio Code here. Then we're going to click the plus sign. type in a prompt. So for example, we've got Mistral div de Devstral selected here. Then we're going to hit enter. And it's usually really fast to reply. So I'll just wait for that to come back. Yeah, there we go. And it's all right. Here's my plan. Here's what we're going to do. So create a new directory. create HTML, add JavaScript, add some basic styling. Boom, shakalaka, let's go. All right. And then it's coding out right here. Now, by the way, if you want like courses on how to set up Visual Studio Code and all that sort of stuff, we've got courses inside the AI success lab. So, you can get all the links right there along with lots of other AI training as you can see. But yeah, basically, it's really easy and simple to build stuff out with this directly. You see how quick it is. It's already created the random password generator. If we run that command, we can start generating passwords. As you can see, how quick and easy was that? And it actually works. Now, obviously, you can improve the URL and CSS and style and everything else. But how incredibly easy was that to do? Right now, if you want to make it look

better, then for example, we can say, okay, make the UI look more interesting, beautiful, and fun. And you can just go back and forth with the AI to make it as good as you want it to be. Right? So if you're thinking, "Ah, this doesn't look like the best thing in the world," it's okay because what you can easily do is you can just give it ideas and then decide what you want to do step by step. Right? You can see now it's going back and forth. It's editing the CSS and the HTML and then from there we get better outputs. The other thing you could do as well is you could switch to an API like Ford just for the front end just for improving the front end and that sort of thing. Now if we run that and we open this up. So you saw the output earlier. This is the new version. Looks a lot nicer. And again, how easy was that? Right? Literally all we did was just say make it more beautiful, make it more interesting, etc. Right? So if you're thinking, "Oh yeah, it doesn't look as good as I want it to. " That's on you to change, right? go back and forth with Visual Studio Code and keep improving step by step. And you see here, it cost us $0. 0. So, it's completely free to build with. And congratulations, my friend. You are now a coder and you don't know how to code. It's crazy. So, let's try another

one right here. And so, if we go over to Devstra, let's see what else we can build. I'm just going to have a look through these 100 ideas. So, let's say create a Tetris clone with falling blocks. Let's try that one. And there we go. All right, you're good to go. Savv says, "Hey Julian, I don't have a site yet, but I'm a software engineer and love your SEO content. Is there a way to access it without needing a website? "

Yeah, you don't need a website to publish SEO content. You can publish content on Medium or LinkedIn or Twitter. That's what we do every single day as well as our websites. But also, if you want to learn this stuff, you can join the AI profit boardroom. And inside there, this is where we teach all this stuff. And honestly, the way the internet is going, if you don't want to, you don't have to build a website, right? There are many ways to get around this. I showed loads of different workflows inside the AI profit boardroom. So, you can just steal this stuff and ideas from me and then get the workflows inside there. So, you don't need a website to be able to do it. And Chad says, "The more these AI APIs and that sort of thing pop up, I'm overwhelmed and I don't know which is better for what. " Right? So, if you want to learn, okay, what's better and what works and that sort of thing, check out the air profit border. And inside there, what I actually do is I do a recap each week, right? So I showed the latest stuff and I show what's working, what's good, what's not, and what is the best stuff that came out this week, right? And then you can filter through all the AI slot and you just focus on the stuff that's the best and the most relevant, right? So that's the way that I would recommend doing it is you just get the updates from the AI profit boardroom. You save yourself literally probably, I don't know, 30 or 40 hours of research and you just get someone like me to filter it all down for you and tell you what's the best for what right now. And also, you can post in the community and ask any questions you have if you're looking for like a specific type of workflow. So, let's wait for this to build out. Right now, this is a bit more complex. It's going to take a little bit more time, but you can say we're going to add colors to the Troino shapes over here. Now, we're building this out step by step. Here's what we're going to do, right? So, it's really easy and simple to build this out. And then we can start working our magic. All right. Now, in the meantime, I'm just going to open up another tab, right? So, a new window. So, you can see this pretty cool. So, inside here, for example, we can actually go over to client and client

have this MCP marketplace, which means you can connect the devstri API to free MCPs, for example, like firecrawl and set those up. So for example, I actually set up the browser MCP as you can see and then you can easily set up MCPs inside client and get those set up and working nicely. All right. So if you want to use like free MCPS along with Devstral and then connected to something like Via Cruel for example, you can easily install this and it doesn't take long at all. But it will make the API way stronger, way more powerful because now it can for example browse the web or it can control your browser whenever you want. Now what you can also do you don't have to code with it. So if you do want to use it in a different way what you can actually do is you can go over to open router and then you can chat inside the free model here. So you can use devstral that way as well. If we go into the chat we can actually connect this to web search as well and then inside the chat here we can say okay let's see what we got. Build a productivity calculator and it will start coding out right there. As you can see, it's super fast to build out stuff if you don't want it in Python as well. So, like for me, for example, I don't want to build stuff in Python. So, I'm going to say build it in HTML. And then what we can actually do from here is we can queue up liveweave. com to preview the code. So, for example, if we take this we've just coded right here, then we're going to go over to here and paste that in. And boom, productivity calculator is ready to go. If you want to improve that, if you want to make it look better on the design and everything like this, I'm just showing you for speed on a video, but if you want to make it look better, just go back and forth in the chat and ask it to look better. Like it's as simple as that. Really easy and simple. The other way that you can use this actually is you can actually go over to Lama and Oama have dev stra available to run offline and you can see it's the best open source model for coding agents as they described. You've got these two options for building this stuff and you've got all the benchmarks here as well. So it's compared to for example Quen 3, Gemma 3, Deep Seek and Devstral is miles ahead of all of these, right? It's really good for agentic coding. It's quite lightweight 128k context window which is quite small but you are running it locally so you know not a big issue there and you can see how it performs on the benchmark. So versus for example GPT 4. 1 mini devstra is actually outperforming it. Same for claw 3. 5 swift etc which I think is wind surfs model and that's really easy. And then what you can actually do is if you download for free so you can download for free there. Then you can go into your terminal make sure that you have running in the background. So for example here you can see is running. And then if we go to Alama and we go to terminal, we'll copy this command. Go to terminal and you can start installing that as well, right? And then you can run devstral locally. So it's pulling the manifest and then downloading it. Pretty simple and easy, right? I'm going to close that cuz I don't actually want to download it right now, but you get the point. Like you can run it locally. You can cope with it for free. It's easy and simple to use. It's pretty powerful and it's really cool. All right, so let's run this command. Now we've got Tedris ready to go and that is working perfectly as you can see. Let's see if we can get this working. Boom. Shakalaki. Yes, it is. All right. Before I get addicted to Tetris, I'm going to close that. Now, also the other thing that I notic that's really cool with Devstral, for example, is if we go back to here, not only can we code for it for free and build some pretty nice stuff, but we can also, for example, ask it to build stuff and it will actually ask us a lot of questions, right? So, if we go inside client here and we say build out an AI growth calculator, you can see it asked me like loads of questions before I actually built it out. So, it wanted like the specifics and it was really trying to plan out the process before it went ahead and built that. So, if we go inside a new chat here, click the plus sign, we'll try the same task again. And there we go. Sometimes the API does fail, but I think that's because we're using the free version. So, you can always wait for it to load or you can run it locally. So for example, if you go to your settings and then select O Lama in the settings, you can run this for free locally, which is pretty easy as well. So you can just choose Lama, choose a base URL, the model ID, and then from there start building out with Olama, which won't have any rate limits, right? So you've got that choice between open router and the normal version. So that's basically

it for Devstral. You can see how we've basically built anything that we want using the free API key. You can use it inside Visual Studio Code. O Lama. Open router chat as well is another good option. I've also created a cheat sheet inside the AI success lab where you can see exactly how to use it, what it means, etc. how it works and how it performs versus other API keys. So, versus GPT 4. 1, DeepSeek and previous open source bests. And then also answers like some of the latest FAQs on this stuff. And it has a quick start guide as well, plus some use cases. Right inside the ASXS lab as well, you've also got 100 fun and simple Devstral project ideas. So, if you want to test building stuff out with this quickly, you can just steal some of these ideas from me. And additionally, inside the courses section here, you can see a bunch of courses. You've got options on how to install this stuff, plus all the links to all the stuff I talked about. So, feel free to get that inside the AI success lab in the comments description, completely free. And if you want to get coaching support, help from me, feel free to get that inside the AI profit boardroom. This is the place where I hang out. You can ask any questions to me, connect with me directly, etc. And also inside the community, you can post any questions you have. So, if you're struggling with anything, feel free to post any questions. We're happy to help you. So, for example, this morning I posted a recap on all the latest updates. You can also see, for example, Aaron posted a question. We got back to him straight away. We actually have this thread right here where we get all the ideas of automations people want to build from the community and then I record a video each week where I make all of those automations and just basically build any automation for you just to help you and show you exactly how to implement it. So feel free to get that inside the AI profit boardroom. Also comes with all of my best agents workflows, templates, tips, etc. and weekly coaching calls so you can ask questions live on the course as well. Link in the comments description. Feel free to get that to go and build this stuff and implement it for you. Then feel free to get that inside the this page right here, link in the comments description. Basically, you can book in a free AI strategy session. What we're going to do is we're going to look at where you spend your time, figure out the best systems to build, and then we can just go off and build it for you if you want, if that's what you want, right? So, feel free to book in a free AI strategy session about that. I appreciate you watching.