China’s Kimi K2.7 is ABSURD!

China's KimiK217 is scary good, so I want to show you exactly why that is. Three days ago, Kimi just dropped from a really small lab in China called Moonshot AI and it's building some absolutely amazing stuff. So it came out on June the 12th and it costs almost nothing to run and this thing just came from a tiny lab. It's open for anyone to use, so it is open source, which means that you can start running it for free.

It's already available on Ollama and I'm going to show you what it does, what it can't do and what this means for you and your business because there's a way to use this right now most people are going to miss and let me show you the before first right. So the old way is you get an idea, you think about building it and it just kind of disappears or you build it and you never find it again. What this is doing is building some amazing stuff and it was super simple. So for example here we have a full agent operating system.

This is literally an operating system with different apps. So if we type something into that it's just going to automatically save. We can close it, open up again and it creates literally an operating system. It's like building windows but we just did it with Kimi K2.7.

Obviously not as complex but you get the point. Here's another example. It can create some like pretty cool animations and stuff. We actually created a full game as you can see right here.

Here's another game that we created and we even created like an open world game as you can see right here right and so you can build some absolutely amazing stuff with this. It's pretty wild how good it is and you can do it in one sentence. Now if we have a look at the benchmarks here we can see Kimi K2.7 measured is a 1 trillion parameter mixture of experts model, only 32 billion active parameter per token and with a 256k context window. It's pretty small actually but it's built for long agentic coding tasks which means like when we plugged it into the agent operating system it could build some awesome stuff simply because we could have it coding literally for like an hour just on its own just getting stuff done.

For example it coded this out. Everything that you see in here it literally coded that itself which is pretty insane when you think about it and it did this all in like I would say a couple of hours. It got all of this done in parallel. Even for example created a full video that's fully edited with an AI avatar and you can see that right here and one of the biggest differences between for example something like Claude and Kimi K2.7 is that this has a coding planned so you can actually plug it directly into Minix.

By the way that is not me. The voice on that, not me. The editing on that, not me right. The scripts on that wasn't me.

I literally just gave Kimi K2.7 a prompt and the skill to create videos and boom shakalaka we could just go off and create all this awesome stuff right and then we've got it plugged into our CLI over here and we already have it inside Hermes. So this is a really powerful model pretty cheap to get. I think I'm on like the $99 a month plan and have come nowhere near using all the tokens. You can see how it performs on benchmarks and it's pretty powerful stuff.

Now GLM 5.2 just came out recently which one do I prefer? Bear in mind they're both from China. I actually prefer using GLM 5.2 for coding in general. I think it's more it feels more like Claude to be honest with you but when it comes to gentic stuff I found the Kimi K2.7 replies a lot faster when it comes to using something like Hermes and that sort of thing.

So that's just one thing to bear in mind as well. But test out for yourself see what you think. And so nowadays you know you can just describe what you want, you get it built, it checks itself and then you get the gallery and this is something else that we did recently is that when we're using something like Kimi K2.5 we have different Kanban boards for different projects. So for example like this one for building out a dragon game and then we have different profiles for creating each part of the report and not only that if you want good quality stuff and I recommend like having this in place is like have a Kanban board or a team of agents and then make sure that you actually have someone who checks the quality of the work and if it's not good enough they send it back.

So here's an example the video judge right here they iterate until the video itself is scored by the judge as above 8 out of 10 and if it doesn't hit that we're not going to use it and it just keeps going back round and round and round until we eventually get the video that we want to view. And so that was how we got that video before created and again you can see on the Kanban board here in terms how we created it and it's pretty nice on the editing it's probably better than most my team could do at this point to be honest with you and this was fully automated with Kimi K217. So if you've not tested it definitely recommend that pretty powerful stuff. You can describe what you want in plain English you don't need to be a coder don't be put off by the name code it just creates code doesn't mean you need to be a coder or technical at all to use this stuff.

Also it can write the whole working file itself you can check its own work and then it lands inside what we have is an agent operating system. I would recommend that too because if you don't have a system to put the stuff in you will lose it. I know because I was just losing all my work for like a year simply because I just didn't have it stored in a proper place so definitely recommend that as well but yeah you get the point like it's pretty cool what you can build with this I mean these are four world open games as you can see and it's a lot of fun to build with it as well you know you create some awesome stuff. Also I think since the whole Fable situation and that getting banned people are looking for alternatives and it's funny the timing of this dropping I think they thought right perfect time to drop this out now and and start getting other people using it.

So when you're at the point where you can build like an operating system and all this cool stuff you this is powerful this is pretty powerful. So the way that it works is you just open up Kimmy Code and you chat with it just go inside the chat then you can write the whole app you can open up the workspace and it's already running you don't need to be a coder you just type in plain English and that's pretty much it. Now there's actually one test where KimmyK2.7 beat Claude Opus 4.8. Now Claude is one of the most respected AR models out there made by one of the biggest names in the space and on this one test the cheap over model from Beijing came out ahead.

The test is about something called tool use in plain words so it measures whether the AI can actually use other tools to get a job done like for example open a file make a change check a box update a ticket the boring stuff the real work is made of and Kimmy scored 81.1 on that test Claude scored 76.4 and the reason for that one matters out of all of the tests that is a tool you use when you need an AI that actually builds stuff so for example apps games tools websites energetic stuff as well now most of the benchmarks they came from Moonshot themselves the company that made it so you should treat them the way you treat any company grading its own homework take them as direction not gospel and independent testers have already started poking holes in them I think one researcher actually ran it against the older version and against another top model on a hard task his take was honestly said the new Kimmy is more honest in how it writes code but on his tests not actually more capable on one job the new version actually scored worse than the old one so it really depends who you ask but from my test it's pretty cool it's a lot of fun to use and it's just another way of you know creating awesome stuff never mind like you can have all your agents just working in parallel as well so here's a real picture Kimmy K217 is genuinely good at building complete things fast and cheap on the hardest most complex jobs the expensive closed models still lead both things are true at the same time why am I telling you the downside because I'd rather you trust me than hype you and because for what you'd actually use this for like for example simple tools forms a tracker little app for your customers the cheap fast good enough option is the one that wins you don't need the Ferrari to drive to the shops now let me give you a sense of how fast this is all moving the same lab launched their first big model in July of last year since then they've shipped five major versions five in under a year the first one then the thinking version then another then another now this one as well and that's the speed this whole space is moving at right here and the gap between an AI can sort this and an AI does this for 95 cents used to be years now it's months sometimes weeks and that speed is the thing to pay attention to not the one model the curve so what do you actually do with all this let me keep it simple because the examples are where it gets real so you could run a small let's say for example you run a small shop you could describe a simple stock tracker and have it built before launch or you could create a website or a landing page let's say for example you sell things online where you could spin up a little tool that tells customers their delivery date one sentence each now I would say the real power of Kimmy is not actually Kimmy itself is having all of them in one place working together and that's what we've built inside the AR platform with the agent operating system it's one dashboard when you plug in all your tools Kimmy, Claude, Hermes, OpenClaude and they all share one memory and one screen so instead of jumping between 10 different tabs you've got one command center you can tell it what you want it builds it and it lands inside your workspace you can even build a full company of AI agents with Kimmy and Hermes and whatever you want inside Paperclip and I've added tools in there for making videos for getting found on Google for everything I do day-to-day and we don't just hand you the tool you get the full zip file a 30-day roadmap that walks you through setting up step-by-step video tutorials and daily updates as we make it better plus four weekly coaching calls where we build this stuff with you live already helped loads of members build their own agent operating system people have never coded a day in their life so if you want to stop having ideas and start having tools that's where it lives link in the comments description or go to the aiprofitborn.com China's KimmyK2.7 is scary good not because it's perfect it isn't but because something this powerful is now open source you can run it for pennies if you get the coding plan and it can build amazing stuff in one sentence the people who start describing their ideas into reality today are the ones who will win pick one idea build it today I'll see in the next one cheers bye-bye