Ornith-1.0 is INSANE (FREE + Local + Open Source)!

Key Takeaways

So, today we're going to be looking at a new local model, Or Nif. Or Nif 1.09B. And I want to walk you through exactly what I've built with it, how it works, etc., how it performs. And this has actually been pretty good, pretty responsive so far. It's It's doing pretty well on the benchmarks. So, you see it's compared against Qwen 3.5, so it's not like the newest version or the the highest frontier level of AI, but for AI model that runs locally, this is one of the best I've seen so far. And these local models are gradually getting better. So, I want to walk you through the whole system and you know, you can run this on your Mac, you can run this on your PC. I'm going to walk you through exactly how it works step by step. We've already plugged it into our agent operating system. So, if we go to the local section down here, we have Or Nif plugged into our local engine. This means we can run AI agents for free, we can run them locally, we can orchestrate them, we can build anything that we want, and we can also preview them directly inside the chat. So, if we go to the preview section here, here's something that we just literally built with Or Nif, and it's pretty fast and it's pretty responsive. Now, if you want to know okay, what's my setup? So, I'm running this on a Mac Studio. Honestly, when I run like even Gemma 4, if I run Gemma 4 on a Mac Studio, not very good. This is uh much faster and much more reliable. So, if you go inside the chat here, um we're like, "Okay, build out a habit tracker." For example, like this, and we can just give it a quick task. It's now going to start thinking locally using Or Nif 9B. And the cool thing about this is well, that it's running locally, it's free to use, it's easy to set up, etc. And then what it does is it will plan it out. So, you can see here it's going to start planning out the habit tracker, and then it's going to start coding it out, right? So, it's going to write in HTML. Now, what we can do from there, once that's finished, is we'll actually be able to preview it and it will be built into my workspace. So, everything that we've previously created with local models, we've got inside this list, we can open them up, we can come back to them, We can open them up in in a full tab, we can get the HTML for each creation. We can preview new stuff here. And then we've got the build section here, and it's just building out. It seems very responsive, and it actually seems to create decent stuff, which is good as well. So, if we look at the benchmarks, it is out So, it's outperforming Quen 3.5 on some benchmarks as well, which is pretty impressive in itself. So, if we have a look here, this is Quen 3.59B, which is down here, and it's nowhere near, right? Then we have Quen 3.535B. Now, this is a 9B model. So, it's a lot smaller, but it's outperforming models that are 35B, which is pretty impressive. It's also outperforming by quite a long way Gemma 4 on benchmarks. So, overall, this is really impressive as a model. So, if you look at Terminal Bench on if versus Quen 3.535B, it's outperforming Quen 3.59B, nowhere near. Gemma 4 12B, which just got dropped recently, nowhere near. And then Gemma 4 31B is on a similar level, but still not as good. And so, on these benchmarks, it's doing pretty well as well. But I always like to test out myself and just see how it performs, and you can see that it's actually doing a a pretty good job. It's It's pretty fast when we use it, and it's easy to use. We'll come back to what we created in a second, and I'll show you what we've built. So, it's kind of like it's near frontier level. I would say the newest version of Quen is 3.7, but this works really fast. It's free, it's private, it's easy to use. So, if you look at like the normal cloud way, you've got your machine coding out, then it goes to the internet to their servers, and then you also pay per API. With this system, the Pocket Frontier way, which is what I'm going to show you in a second, the model lives here on your machine. Nothing leaves, it works offline. So, if you don't have Wi-Fi, you're on a plane or something, you can still use it. It doesn't cost anything to use, it's free. And then it's not it it's completely like sort of sandbox inside your machine as well, which is great as well. So, we've got a 9.5 GB model building real apps locally, privately, and for free. And nothing leaves the machine. And it can run while the internet's off, it can run on a flight, it still works. And it's performing really well on the benchmarks. So, if we have a look as well, we've now got the preview ready to go. I also like this system because then we can come back to everything we've built, we can see the full conversation history. And if you want to preview what we just built, we can open up like so. We can add a new model here, so test, for example, and that's added inside our habits. We have a weekly completion section, and it's easy to use as well. All right. And then we can also open this up in a new tab. It's all organized, we can download the HTML, and it saves inside our workspace as well. It's pretty cool. Now, I'm not going to say like it's I mean, it's definitely not up there with like Claude Opus 4.8 or something like that, but as a local free model that's fast and responsive and just ready to go whenever I need it, that's pretty cool. And this all plugs into our agent operating system. So, for example, what I could do is I could use that local model, and I could use it with Paperclip and build a whole team of agents. I could use it inside Hermes, and then we could use Hermes directly with this as well. So, we actually set up a separate agent profile for Orni. And this is basically Hermes, but plugged into a local model. So, we can use Hermes with Orni as a local model here as well, which means that you can use it agentically, and you can build cool stuff with it as well. And then you can get free access to it via Hugging Face. You can also use this on LM Studio, too. So, LM Studio is a good way to use it. But yeah, it's basically a a full family of open source LLMs built specifically for agentic coding. Now, you have 9B dense, 31B dense, 35B MoE. So, it's a full range. This is a a tweet by Round Table Spaces, as you can see. And it works with Hermes directly. It's a little bit slow on Hermes, you know, because it is in the Gen 8 model running locally. But, still, it actually works and you can use Hermes Agent for free with it. We built and tested out a bunch of stuff with it just to see what it's like, as you can see. It is pretty cool. Like, you can build some interesting things. You might also say, like, does all this stuff work first time? Most of it works first time. Some of it, like, didn't work first time when we were testing it out. So, that's just something to be aware of, as well. It's like it you know, it's With local models, usually what you're getting is like kind of like almost like last year's model, if that makes sense. So, if you look at the benchmarks of like Frontier level models last year, that's typically what you're getting with local models this year. So, it still means you can build something good, but is it perfect? I wouldn't say so. You know, it's just something that helps you run free models. So, it also depends on your setup, as well. Now, the cool thing is with our local engine over here, is it actually finds its own bugs. So, if we go to the local engine section and we go to the local setup here, basically, when we're building out with this, it'll write the app or whatever we ask it to build with a single prompt. It'll run it, find bugs, and then fix itself. So, you can fix it directly inside there. And the great thing about this is if you look at the old way, you know, everything you were doing goes to a cloud, you pay per API, you get rate limits, it's not private, and if you have no internet, you can't use it. With the new way, the model lives on your machine, so nothing leaves. It's zero cost to use. There's no limits. Your data stays private, and it works with the Wi-Fi off. Now, you might say, okay, why not just use Claude for everything? Honestly, like, Claude is is a great model, and I still do use Claude for most things if I need speed. But, if you're, for example, looking for a free way to use AI agents or for example, you prefer to run privately offline, then this is a better way, right? You can't do that with Claude. So, how does it work? Well, you just get the model, as you can see. Then you can wire it inside the agent operating system. You might also say, "Okay, you know, isn't local setup too technical?" It's just like a couple of commands, right? So, these are the commands that you can run and you can use it with LM Studio and it's free to get on Hugging Face as well. The other cool thing is that we have a lot of systems inside the agent OS that you can plug into this. So, for example, you could use this with the local Hermes engine inside the Hermes tab. So, we've got a separate profile for One If ready to go. Then you could also use it with the agent Kanban. So, the agent Kanban is a local team of agents that you can organize on a board like so. You can see what we've built out as a workspace and this is a board of agents that just work with local models. So, you can orchestrate your agents, have them working together and have them working for free. You could also use them inside the local chat section. And something else that we've set up recently is a loop engine. So, what you could do with this system for example, is you could use a local model like One If as the builder and then you could have like a free API judging the work and loop it round. So, that's another way to use this as well. So, you see how the local model just plugs into many different workflows inside the agent OS. You got Hermes, agent Kanban, local chat, the loop judge and even the video agent would be pretty interesting. It could write the scripts for your videos, too. You might So, I mean some limiting beliefs to cover here that might be holding you back on this. So, number one, people believe like powerful AI has to live in the cloud. You can run this locally and you can actually build stuff. You also might think, "Well, free means a worse model." That depends what you're comparing it to. If you're comparing it to like Fable 5, of course. But if you want something local that runs for free, well, this is a decent option. So if you want to get my setup with the Agent OS and all the systems we built around local AI, you can get that inside the AI Profit Room. Link in the comments and description, or just go to the AI Profit Room.com. Inside the community, you can ask questions, get help and support in real time. Inside the classroom, you can get access to all my best trainings. We've got a lot of trainings on local models. And then you can get the Agent OS over here with a video tutorial on how to set up, a full guide, and step-by-step daily tutorials. Inside the calendar, you can join our weekly coaching calls, ask questions, and get help and support in real time. Inside the map, you can meet people in your local area using local AI. We actually have one member, uh Daniel, who's, for example, like he's got a local setup. So there's a lot of people using local AI inside the community. He actually created a full AI avatar video using local AI, which is pretty insane. So, you know, you can you can get some seriously good setups with local AI if you know how to, and there's a lot of people inside the community you can learn from on how to do that. Thanks for watching. Choose your