So, China just released Kim K2.7 from Kim Moonshot and it's pretty impressive so far. So, this is an example of a game I actually created using Kimik
K2.7 and the new release here and it's available on the coding plan if you haven't got access to it already. It's pretty powerful so far from what
I've used. Now, one of the things I will say is like this was not just like a, you know, one shot and
then we're ready to go sort of setup, but it works pretty nicely. We've created four games with it as you can see right here. actually works. And the
other thing as well is a long horizon agentic model, which means like it can code and build stuff out like you see right here for hours. And there's
some other stuff we built out with this as well. Some of it I like, some of it I don't like. So, for
example, if we open this up, like what is this, you know, like it looks cool, but and I love the colors and everything like that, but is it actually
useful? I wouldn't say so. Um, something else that we actually did as well is we created this fully edited uh video with my AI avatar and it was built
with Kimmy K217. So, what I actually did is a team of AI agents is like edited it, created it. You
see a gen CLI as you can see right here to build this out. Here's another example. We actually built out an operating system. So if we open this up,
right, what we've got here is an operating system where we've got Kimmy OS and then it built out the whole operating system if that makes sense. So
for example, all these apps inside the system are fully created with Kimmy. And the other interesting
thing about all of this is it actually judged its own work and then kept improving from there, which is pretty wild as well. So, if you don't have a
Kimik K217, maybe you're looking for a cheaper alternative to, for example, like the Chat GBT subscription. So far, we've been using it a lot and not
run out of tokens. I'll show you what model I'm on. So, there's several models for Kim K 2.7. You
know, there's several different plans. Let me show you what the plans are in terms of the pricing. So, if you go to Kim K 2.7 here, as you can see,
you can use it on the API. I just use it on the coding plan. Let me find that for you. And we are on this plan right here. So you can see we only use
like 2% and 0% of actual weekly usage. Now I was on the plan that's lower than that. So I started off
on this one and then you know after iterating all that stuff that I just built a second ago, it then we then needed to switch to this plan. But as you
can see right here, the token limit is pretty good and you can build some amazing stuff with it as you've seen today. So Kim K 2.7 is a really cool
model. a lot of fun to build with, a lot of fun to create cool stuff. You see some other stuff that
we built right here. Uh very visual stuff as well, which is pretty cool. We actually tried creating like a Skyrim style game. You see, we called it
the dragon realm. Now, if we open this up, like you can see, we can go through this virtual world. Now, I think if you actually gave it like a skill
on how to create games, it could create something better. But this was just kind of like straight out
of the box and it created a whole sort of virtual world where we can go through it and then we can play games and uh and we can build that out. You
know, we can make it as complex or as interesting as we want. So, pretty amazing stuff, especially, you know, with the news of Mythos and Fable going
down, right? this might be something that you're like, "Okay, well, if I'm going to switch coding
plans or if I need something that's pretty powerful, then you can go with Kimmy K217." So, and you can see some other cool stuff here, like this was
like a dragon racer game, uh, synth wave style Kimmy K217 banner. We got a galaxy that it built out here and some cool orbiting stuff as well. Now,
you can also plug this into Hermes. So, for example, here you can use this as the model. So what that
means essentially is like if you're on the coding plan with Hermes and I'll do a separate sort of tutorial on this. But if you're on the coding plan
with Kimmy K2.7, you can change the model or have separate profiles for Kimmy K2.7 as the API model plugged in. So for example, if we go over here and
we take a look inside here, we've got Kim K27 as a separate agent profile, which is pretty cool cuz
then like I can chat with different models using Hermes across different profiles. So, for example, here we've got N2, which is another model from
China that just dropped. It's a free API. And then we've got Kim K27, and we can see the conversation history for each. Super easy to chat and switch
between them. And then that just makes everything easier, right? So, if we go to Kim K2.7 here, and we
just say, okay, are you here? And I want to see how fast this is actually to reply as well. So, let's see how quick it is to reply. Maybe it's not
that fast. It's like it replies in seven seconds. So, it's not instant. It's not like super fast, but it's pretty good. Pretty good stuff. And then
what we've actually done is you can get Kimmy code again on that CLI coding plan and we've plugged that
into the agent OS here. As you can see, you could use this separately as well. And um and then everything that we create with it, we've got it plugged
in here. You may say as well, you might be looking at these and going, "Okay, well, does that mean Kimmy can actually create images?" So, what we
actually did here is we used chat GPT image 2.0 as an API. gave that to Kim K2.7 code and then it could
just go off and build stuff out as you can see and then you know it can create your usual sort of fun games and that sort of thing as you can see here
as well. Let's have a look what else. Yeah, so the web OS was pretty impressive. This was kind of like a drum studio that was a lot of fun to build
but yeah it's pretty nice and again like if you want an alternative to Claude this is a decent option
like there's nothing wrong with using it at all I think honestly. And then something else that you can do actually I'll come on to that in a second.
But if you go straight to kimmy.com and you're on the coding plan, then you can switch to kimclaw and you can actually have like a version of openclaw
but in the cloud that's cloud hosted and kind of silosed away. So that's quite good because you know
if you're not technical you don't want to set up openclaw yourself or you can use it over here as well which is pretty cool. And then you've got Kimmy
code directly so you can run it inside your terminal. I prefer to use it inside the agent operating system because it just looks a lot nicer and it's
easier to use. But you can run it inside your terminal using this system as you can see here. And
then we got, you know, the agent swarm feature. This was all stuff that was already there inside Kimmy before. You got the deep research docs, the
websites, you could build the slides. So if you go directly to the platform and you're on the plan, it's very useful. You see your chat history as
well. And there's an app, there's a desktop app kind of like claude desktop but for Kimmy as well. Again,
what do I prefer? Personally, I prefer to just have it inside one system because then I can even have like a group chat with Kimmy inside there. I can
have my Hermes agents walking over here, etc. You know, and again, you might say, well, can you use the subscription with Kimmy on Hermes or your AI
agents? Yes, you can. Yes, you can. So, you just plug it in as a model. Um, and what you would do is
you would create an API inside your coding console and then from there you can use it. So, that's what we did. We created we created an account inside
the Kimmy coding plan. Then we got an API inside the console. Then we plugged that into Hermes. You might be saying as well how do you connect it? How
do you get the model and connect it to Hermes agent? So you can add new models inside your
dashboard here which you can see the other option that you have. So you can change the model as you can see here right. Uh the other option that you
have is you can just go straight to terminal like so. Type in Hermes model. Then you would scroll down to Kimmy and then you can switch between these.
And again, we're on coding plans, so that's what we plugged in as the setup. Pretty cool. Yeah. And
the other interesting thing is like you could actually plug it into claw code. So you could use free clawed code, plug in your coding plan into this
as well. And then you could use free clawed code with the coding plan as well. So what that would mean essentially is you have the harness, the agent
harness of claude, but you've plugged in your own API into it, which is pretty interesting and pretty
amazing when you think about it. So those are multiple different ways that you can use this stuff and how you can use it together as you can see. And
yeah, it builds amazing stuff. Did it all build it in one go? No. But is for example Claude going to build this sort of game in one single prompt?
Probably not, right? You have to go back and forth with it a lot. The other thing you can do is you can
actually get the CLI and you can get Claude to speak to the CLI and build cool stuff as well. So that way like it's operating the CLI for you in the
background on long horizon tasks. So if you're wondering what that looks like, well, let me show you how we built this into the agent operating
system. So essentially what you can do is you can have a conversation just for Kimmy K217. You give it
access with the API key to your coding plan and the CLI and then it can actually operate Kimmy coding uh and Kimmy K217 in the background as well,
right? And test out for you and give you screenshots of what's working as well. But I think that's quite a good way because then you've got Claude as
a brain and it's delegating and testing stuff inside Kim K2.7 so that you don't have to do all the
testing and the prompting yourself. That saves a lot of time as well. Now an interesting question that I get all the time is like you know have you
got multi- aent orchestrations running autonomously or you manually orchestrating this stuff? I'll show you an example of how it can work autonomously
as a team of agents. Right? So we got this camb board and we have uh this video editing team of AI
agents as you can see over here. So we're using Kim KC17 as the model to orchestrate agents and then build stuff. So for example, if we plug in a
prompt to generate a video here, we can have a whole team of AI agents that edit and create an avatar video and iterate until it's done. Now what you
actually see is that we have this judge and this is really important to the whole system. You have a
judge AI8 that looks at the whole quality of the output and then tells you if it's good or not and whether it passes and if it doesn't pass then it
will keep iterating and improving it. This is actually something I saw on a recent Nate Herk video. So I was like let's test out the same thing inside
Kimmy K27. So credit hint to him. Like he's he's absolutely awesome at creating videos. But if you
have a look over here, for example, this video that we created, you know, that's fully animated, um, fully set up, edited with Kimmy. It's got audio.
It's got, you created AI music in the background. This whole thing was fully edited with a team of AI agents. And literally, all you have to do is
give it a prompt, and then from there, your AI agents will handle the rest, which is pretty amazing. We
did another example here where we had a team of agents of Hermes agents working together. And you can do this with communicate 2.7. So we created a
page on the hour profit boardroom, gave it a keyword, and then had a team of agents work together and autonomously create this stuff as well. So
that's basically how you can have your AI agents working with Kim K27. So pretty amazing stuff. Definitely
recommend checking out. If you want like a a cheaper alternative with more tokens to Claude, then it's a good option. If you want to get my full
system for the agent operating system, you can get that inside the AR profit warning link in the comments in the description or go to the arr.com.
This is my AI automation community for helping you save time and grow and learn with AI automation. Now, if
you go to the classroom here and then you check out the agent OS system tutorial. We've got the video tutorial on how to set up when it was last
updated and also the resources. You can download the zip file from us as well. And then we have new daily tutorials like you can see on all the cool
stuff that comes out. Also inside the community you can ask questions, get help, get support in real time.
Inside the calendar you can jump on weekly coaching calls. We have four weekly coaching calls where you can ask questions, share your screen etc.
Inside the map you can actually meet people in your local area who are building with Kim K2.7 and other AI agents like Hermes. And this is all inside
the AR prof. OT inside there. Link in the comments descript.