[00:02] explain this using analogy. Say you have your favorite coffee shop in the town, your favorite pizza place, your favorite Punjabi restaurant, and to order a food, you are using their individual application. So, to order coffee, you
[00:17] use the mobile app built by that coffee shop. To order pizza, you use a mobile app built by a pizza shop. Now, this works, but you have to maintain three different logins, you have to install those apps. It's lot of work. Imagine a
[00:31] single application like Zomato is present on your phone, where you can access all your restaurants, and [music] things become so much easy for you. things become so much easy for you. Open router is a Zomato for LLM APIs.
[00:45] It's a unified interface for LLMs, where you not only get access to paid LLMs, but lot of free LLMs as well, which [music] makes things easier just in case if you are a student learning AI. Using these free models, you can get access to
[01:02] [music] these API calls, build your AI projects. And if you're doing any professional work, in that case, you get access to your commercial models as well, all in one place. This is their website, and they have so many different
[01:17] models available. So, if you click on this model list, you will find 646 this model list, you will find 646 model. Many of them are actually free.
[01:29] So, if you search for free here, you get access to all these free models. And if you're building any AI project, any hobby project for exploration, etc.,
[01:41] using this free models will be helpful, because then you save a lot of money on the cost for API tokens. Many companies also use this for their beta testing
[01:53] ground. So, if you search using alpha, this model, for example, Hunter Alpha is this model, for example, Hunter Alpha is next gen model from one of the prominent companies. And if you look at their provider here, it says stealth, meaning
[02:09] provider here, it says stealth, meaning this can be OpenAI, this can be X, you know, XAI. When these labs are releasing new models, they will first push it on Open Router, some of these labs, and they
[02:22] will get the beta testing or alpha testing here, and then eventually they will unveil that Hunter Alpha was nothing but GPT-6. So, let's go ahead and get the API key. So, click on here, and you can log in using your
[02:38] Google credential, GitHub, or create a new account. All right? Now, here, you new account. All right? Now, here, you can go ahead and create your API key. Once you create this API key, how do you access it? So, click on documentation,
[02:52] okay? So, see, you can access it using OpenAI model. So, now, in Python, you can import the OpenAI model, and then in the base URL, you can say Open Router, and then your
[03:10] API key here. Same way, if you're making a a, you know, HTTP request, you pass URL here and your headers, etc., where you will specify all these parameters, and in here, you specify your model, etc. So, this way, you can specify any
[03:27] model, by the way. See, all these models. You have so many different models here. Now, let's say I'm using this XAI Grok model, okay? So, you can click on it, and it will show you some statistics on this model. You know, this
[03:42] is the throughput, latency, etc., and it will also show you the applications which are using this particular model. For example, this application, Kilocore, used 724 million tokens, okay? Uh you can see all
[04:00] other statistics, etc., and you can get access to the code as well. So, let's say you want to call this model using Open Router, in that case, you will just say request.post, okay? This is the model,
[04:15] and that's it. Like here, you are specifying the API key, obviously. Now, the way pricing works is, for this model, for example, if you go to pricing model, for example, if you go to pricing tab, it will charge you $0.3 for 1
[04:27] tab, it will charge you $0.3 for 1 million tokens, okay? So, let's say I'm building my AI app, okay? And that app uses, And that app uses, let's assume Llama model for some low
[04:41] cognitive task, you know, for some simple task, it is using Llama model, okay? It is also using Claude code, you know, Claude, not Claude code, but some Claude, let's say
[04:53] Opus model for high reasoning task. Now, in a typical architecture, you will have .env file, where you will specify have .env file, where you will specify your Anthropic key, right? Anthropic
[05:07] key is equal to something, then your Llama key is equal to something else, etc. But with Open Router, what will happen is you have your AI app, and all you have is a
[05:23] your AI app, and all you have is a single unified interface of Open Router single unified interface of Open Router in your .env file, you have Open Router key. Okay, you have just single key, and then
[05:36] call, you can make call to variety of models, you know, Llama, Claude models, GPT models. In your request, you will just specify single key and say, "Okay, this is the model I want to call." If you
[05:51] want to test capabilities of any model in a playground, then you click on chat, and here, you will interact with it so that you can understand, you know, what that model is capable of. Not only that, you can
[06:05] show you. See, when I get this, I can add multiple models. So, I'm going to, let's say, use best coding models here, okay? Best coding model, and I will say, "Write a program in Python." Let's say
[06:23] I'm writing this kind of uh program. Of course, I need to buy credits, or I can continue with the free models, okay? So, let me just uh open the free models Let's say this is the program I wanted to write, and I have Hunter Alpha,
[06:39] Healer Alpha, and both of them are now writing the code. I can check the quality of the code produced by both the models. So, see, Hunter Alpha produced this particular code. I can take a look at it
[06:53] uh along with some uh explanation. And my Healer Alpha is I think it is still producing. It is taking more time. I mean, based on this, clearly I like the Hunter Alpha. Healer Alpha is kind of lost. So, you
[07:09] see, this way, you can uh compare different models. You can check rankings as well. So, see, these are the top models based on weekly usage of models models based on weekly usage of models across Open Router. So, if you look at
[07:24] this This is like daily, right? So, daily or weekly, I think. Uh if you look at March 9th statistics, MiniMax, which is I think Chinese model, is 1.75 trillion, which means all the users on Open Router platform, in in aggregate,
[07:40] Open Router platform, in in aggregate, they have used 1.75 trillion tokens for MiniMax model. Then, Step 3.5, Deep Seek. These Chinese models are cheap, by the way. Some of them are free, as well. Then, you have Gemini, Claude Sonnet,
[07:53] Claude Opus, and so on. And here is the LLM leaderboard, okay? This is comparing the most popular models on and obviously, number one is MiniMax,
[08:05] based on the tokens usage, okay? You can see this way, this week, or this month. See? Uh for this month, number two is Gemini 3 Flash. And they have benchmarks, as well, you know? They have some kind of
[08:19] intelligence index score, coding index score, and so on. And based on that, you can check this scatter plot and find out which model works the best for your needs. And there is some comparison on latency, as well. Now, if you go to
[08:34] latency, as well. Now, if you go to apps, uh you will find a couple of apps here, and the way it works is this largest public apps, such as Open Claw, they will opt into usage tracking on Open Router, okay? They have opted into
[08:49] that, and based on that, Open Claw has already used 11.1 trillion tokens, okay? T means trillion tokens, and these are the trending apps. Of course, Open Claw and Claude code are the biggest ones. Now, let's say you are
[09:04] using a framework such as LangChain, then also you can get Open Router integration. So, you go to docs, and here, in the third-party integrations, a code like this. So, from LangChain Open Router, you can import this,
[09:20] and once again, here you specify the model and your temperature. And of course, you will be providing your Open Router API key. So, see, they have Router API key. So, see, they have integrations with LiveKit, LangFuse, all
[09:33] eventually, on your dashboard, you will see how much you have spent, request, your tokens. See, we used Healer Alpha, Hunter Alpha, so there are many tokens. You will see the logs, credits, and so on. All right, folks,
[09:48] that's it. I hope you got some understanding of Open Router platform. The free models, especially, are going to be very useful if you're learning AI, if you're building AI projects. Thank you for watching. If you have any
[10:01] you for watching. If you have any questions, post in a comment box below.