OmniRoute: Run AI FREE Forever!

**Julian Goldie** (0:00)
Hello, I want to show you a powerful new tool that means you never have to pay for AI again. And this is a really interesting, new, free, open source project that can work across all of your favorite coding tools. It's called OmniRoute. And what it means essentially is that you can use it in any way you want, whether that's directly with your tools, whether that's, for example, inside your agent operating system, you can build whatever you want. And then also, when you're doing this, you never have to pay for AI again. And I'll show you why in a second, but let's just demo it out right here. So, for example, if we say, okay, create a galaxy-themed website with amazing animations on a simple landing page that looks super domain-inducing and fun, right? It's just something basic like that. So you quickly type in the prompt here, as you can see, or you can use Whisperflow, we prefer. And then what you're going to see is that you get the prompt here, then you're going to see that it starts routing through free providers. You can see what you've saved, so what you previously created down here. And then you can see the preview of everything you've built and the code over here. And you can save and download this whenever you want. So it's a really powerful tool where you can code for free, preview it, save it for later, customize it however you want. And it's free. In fact, this means that you never need to stop coding again. So, you know, if you're worried about tokens, if you're worried about using agentic tools, if you're worried about, like, your builds just adding up for APIs, I don't see why you wouldn't use this. It's pretty unbelievable. And so the way that works is basically you connect every AI tool to 237 providers. And over 90 of them are actually free through one single endpoint. Now you could use this agentically as well, because you can use custom endpoints with something like Hermes and then plug in free APIs into Hermes as well. Now you can see that we get the code back directly over here, directly from our setup, which is pretty cool. And if you want to see some stuff we already built previously, you can see an example right here. We built our animated keyboard. We created, for example, a to-do list app and also a Galaxy background, as you can see. And then if we want to see the code, we've got that over here. If we want to save it, we can save it for later. And we can open that up in a new tab as well. Be amazing stuff. Now, also, what's really interesting about this is it uses two token minimization playbooks. So number one is RTK and number two is cave map. So what that means essentially is that when you're coding with either of these tools, you save 15 to 95 percent of your tokens, which means that you're never having to worry about rate limits because let's say, for example, you've got one endpoint, so you don't need to configure anything really once you set up. Then you have 93 tools that it runs through and they connect all to one endpoint. You can plug it into Claude Code. So Claude Code, Codex, Cursor, Client, all these can be used for free as well. It's got auto fallback. So what that means is if one of these 90 fails, no problem, you can plug it into another one. Let's just try something else out now. So I'm going to say, okay, create a habit tracker app like so.
I'm also waiting for that.
What that means is like because you're rooting through so many different APIs and you're limiting the number of tokens you use, you're not going to hit rate limits, so you're not going to be kicked off using that particular tool.
Usually, for example, if you're using free APIs or free tools, you can get rate limited very quickly, which means that you've run out of tokens, your tools just stop working. You're like, it was free, but it doesn't work. That has often happened to me in the past. And it's not like I'm a miser in front of that. Like I really love using Claude. We use Fable, Fable 5 all the time. I mean, I actually have my own benchmark website where I test 45 different projects across all of my favorite models. And you can see an example of that right here.
You know, for example, like with Fusion, and we create amazing stuff, right? I'm more than happy to pay and try different things. But my point is, if you're worried about free APIs, which I know loads of people are, then you can use something like this to help you. That's what we want. I think that's fantastic. Because it means that AI is free for everyone. Also, this is open source. So it's available on MIT license, which is great. And then the promises here are that you never hit limits. You save up to 95% of tokens. It's free to start. There's a free tier.