today I'm going to show you how you can use a new free AI coding assistant in vs code and keep your Code 100% local and you can use free open source models that are capable of giving you as good or even better results than you may be getting from a co-pilot subscription so if an AI code assistant can deliver this much for free its earned instant Chad status I want to guide you through this in 10 minutes or less so let's get right to it we need to start today at ama.com now note you need to download this for whatever op operating system you have it's available for Mac Linux or Windows Windows is a preview so when I click on the download link you can see it says Windows preview that's what I'm using today but for Mac OS it's not a preview and for Linux it's ready to go too so whichever you need download and install that now okay you should now have Ama downloaded and installed now I've got VSS code open right now and I'm going to open up a terminal window with control in the back tick now I will tell you on Windows since it's a preview I had to restart my Windows machine before it would recognize AMA at the command line so you may have to do that if you're on Windows as well now just to verify your installation type O llama and then d-el and it should recognize it if it doesn't or say it says it can't find the word olama restart your computer and then try it once again okay you should have olama installed and you verified that now I'm looking at the eval plus leaderboard here and you can see base tests on the right and the eval plus tests on the left and these are for the free open- source models and some paid models as well it's comparing all of them we can see here just looking in the top five there's code Quinn at number five just above that here in the eval plus test is gp4 and gp4 turbo then there's deep seat coder V2 and then finally another gp4 turbo as of April 2024 here at the very top now I'm making this video in June so I don't know the next time they'll update this leaderboard but you can keep check in back now I'm going to use code Quinn it's very close to these others you could also use deep seek coder VT V2 as well or you could use these paid models but I'm going to use code Quinn and I'll show you why let's go back to oama on that page and we can search models here so I'm going to search for code Quinn and we should be able to pull that up there is Quinn 2 that's not what I want find code Quin there it is and now that we've got code Quin pulled up notice the size over here it's 4.2 gigs that's a pretty good size download but if we go back and say we wanted to go up to that other one that was ranked a little higher deep seek coder dv2 let's go ahead and click that now we can see it's 8.9 gigs so I went with the smaller one and I've been happy with that of course you can use any model that you want to so right now I'll just show you code Quin once again and I'll click that and here it gives you the command to run at the command line oama run code Quin so you can just copy that go back to your terminal in vs code and run that with olama it will take a while to download because it's not that small it's 4.2 gigs but still half the size of the deep seek version that was also available let's make the tutorial official and roll that [Music] footage and now we're going to add the continue extension to vs code this is at continue. deev but we'll just search it up in vs code you can see it's also available for jet brains but it is the leading open source AI code assistant and you can use these free open- source models with it okay we're back in VSS code now let's go ahead and click on the extensions icon over here and when you do that you want to search for continue now we'll pull up the continue extension once we click on that now I've had it installed before so I hope it has the same behavior when I reinstall it I'll go ahead and install that now it shouldn't take too long and once I have that let's go ahead and close that out and now we'll hide that as well and we should be able to open continue from this menu if you have a little more room you may see the icon I'm just going to click continue here from the three dots and it's opened up a chat session so you may not see that at first I am not for sure because what I saw at first was a splash screen let me click this yeah it's not giving me the splash screen it's just giving me the continue so we have Ama installed it's going to have a couple of recommendations for you on the splash screen that we actually don't want so you can go ahead and Skip those what you do want to do is go ahead and click the configure continue to bring up your Json config now inside of here you notice it's got models listed it defaults to llama 3 but then over here when we have our window open you can choose code Quin that's what we want here and it just that's under the auto detect then if I go ahead and hide that window as well you can see here it's got custom commands I'll press altz to wrap that down so there's a custom command in here for test if we type test into the chat it's instantly going to use this prompt with the AI model and when you come down to tab autocomplete model at first yours won't say code Quin in here so notice I've put in code Quinn for the title and code Quinn for the model so it's going to use code Quinn for the tab auto complete let's go ahead and add one more custom command so I'm just going to select this and do shift alt in the down arrow and then I need to put a comma after the first custom command I'm going to call this one step and I'm going to say uh explain the code that is selected here you have the selected code I'll say explain the selected code step by step that's all I really want to do so it gives me steps if I want it to explain anything and then the description is I'll just say code explanation and that's all I need so all I'll need to do is highlight some code and put step into the chat and it will explain it oh I should get rid of my typo explain the selected code step by step there we go now let's try all of this out in the chat window I'm going to quickly open the file tree over here now I've got a python project but I'm going to create a Javascript file at first I'll say convert to my SQL Tim stamp. JS so kind of a specific title there but now I'll open the chat window by clicking the continue icon or if it's not there click the three dots and find it and now I'm going to give it some Specific Instructions here inside of this chat I'll say well what I can do is actually it says control I to generate code so let's hide that and I've got control I now I'm on Windows it might be command I but here I can write the instructions so I'm going to say write a function that accepts a JavaScript date as a parameter and returns a my SQL timestamp press enter let's see what it generates here is the code and it's generating that out piece by piece here now that looks pretty good so we could chat about this code as well so this is the code I could highlight this and now I can do contr L command L if you're on Mac or Linux and it's got all of the code in here and now I can say would this be more efficient if you used two ISO string see what it says here after we get our reply it's taking just a moment and I'll scroll down here it all is says no it would not necessarily be more efficient it gives a full explanation here it says using two ISO string would also convert the date into a different string format it has the T and the Z but it's less flexible so a good conversation there chatting about the code itself so now I'll say can you write the function with to ISO string so here's the example now it's going to give us a different fun function than what we had over here and let's say we like that after it gives us the full description so we can come back up here to the top of this and says apply to current file I'll just click that and now it's putting that over here instead of the code we had and if I quickly hide the chat window we can see it a little bit better but here is the new function now I'd have to go through it also gave some example usage at the bottom of course you don't want to just trust AI you want to check it all out and see maybe you could do something different okay I've got a new python file now and I'm going to use control I once again and let's see what it can really do so I'll say write a python rest API using fast API and SQL Alchemy and let's see what it generates now I may not stick around to show all of this but it's interesting that it might be able to crank out an entire program for us and then of course we could look through it and see what we like what we want to change but it sure seems helpful and it seems like it's every bit as efficient as GitHub co-pilot hey I hope that helped you guys set this all up I'm going to write a blog post about this too and I'll leave links to everything about this video tutorial in the description most of my videos are based on your requests so be sure to leave a comment let me know what you want to see in my upcoming videos a quick thank you to my patrons holy coder is a progress provider L dad is a senior and my junior patrons programming polyglot Tim Philipe Morgan Isaac will Ernie Scott Stacy Philips Abe Javier Michael Alexi you're helping me reach my goals thank you and if you haven't checked out my patreon it has exclusive content and early release content and it's not one of those patreons that doesn't get many posts I'm active on there every week I'll put a link to it in the video description so please check it out if you haven't remember to keep striving for Progress over Perfection and a little progress every day will go a very long way please give this video a like if it's helped you and thank you for watching and subs subscribing you're helping my channel grow have a great day and let's write more code together very soon