I Tested Qwen 3.6 Plus Model… It’s Wild 🔥

Hey, how's it going, guys? In this video, we're going to look at the newly launched model, Qwen 3. 6 plus, by Alibaba. So, if you look at here on my screen, it has launched last week, and this model has been, you know, created for agentic system. That's how it currently looks like. And, uh if you look at, of course, we're going to go through how we're going to use this. And I've already have found a jackpot that we can use that in free via OpenRouter, at least for now. And if you look at here, it says agentic coding front end to repository level, right? Improve multimodal capabilities. So, it also has multimodal capabilities, so we can pass images, you know, we can pass media, and it can also, uh you know, perform reasoning on top of it. And reliable for developers, that's what they say. Performed really well on bunch of, you know, benchmarks, right? So, if you look at terminal bench for coding, software engineering SWE bench flow for coding, uh multilingual, etc., etc. It has performed really well, right? Which is [clears throat] fantastic. Uh if you look at here, the comparison of all these models, starting from Opus 4. 5 to Kimi K 2. 5, GLM 5, and Qwen 3. 6 performed really well, right? And uh has performed well on multiple benchmarks. Even head-to-head to Opus 4. 5, if you see, has performed really well on bench 2. 0, right? Has surpassed Opus 4. 5, which has been treated as one of the best coding models right now, right? So, it's really performing well on the coding agent side. General agent also has performed really well. Uh and knowledge, instruction following, so you can look at all the benchmarks over here, right? For Qwen 3. 6 plus. Now, the good thing is that you can use this model everywhere, right? Okay, like if you want to use this, uh through Open Claw, you can use that. Through Cloud Code, OpenRouter, you can use through uh Model Studios, so on and so forth. So, you can just use through their official SDKs as well, right? So, that's fantastic. Now, they have shown some demos over here that you see how you can use through Open Claw. And if you don't know Open Claw, it's formerly Model Bot and Cloud Bot. Uh you can see it over here. It says a self-hosted open source AI coding agent. And connected to Model Studio, this is how you can do it. You just need the API key, and I'll show you how you can get your API key right here. So, this is a blog. I'll give this link in description of this blog. So, this blogs would be your only source. Of course, there are videos also that you can go through, like including this one, but this blog is has to be your documentation for now. You know, if you want to really use Qwen models. What I have done over here is, you know, I have, of course, built this interface, and I'll give this uh repo through GitHub. I'll give the link in description. Now, if just let's say, uh how do we uh you know, build uh fast API backend. Excuse me. For inferencing an ML model, like let's say if you have trained a machine learning model. Nowadays, we are not doing that because we are all so busy in generative AI. So, let's say once I do it over here. Right now, I'm using through OpenRouter, as you can see over here. And you can also use through DashScope. You basically you need an API key from Model Studio, and I'll show you how you can do it. Now, right now, the beauty of this is that OpenRouter has made it available, right? For free, so you don't have to pay anything, so just for time being, right? And this is where my code is. If you look at here, this is how I have built it. So, this is running through a fast API endpoint. You can see and this has a streaming response. You can find it out over here. Looks so good. Right? So, fantastic, right? I'm using it for free. One of the best models right now, which is available to build code coding agents and even to build general agent, you can do that with Qwen 3. 6 plus model. You can look at this interface that I have created over here. You can find it out over here, right? Such a beautiful answer. Looks good, right? So, if you look at here, it says building a fast API backend for ML inferencing is straightforward, but production readiness requires, of course, careful attention to model loading and async behavior, of course, validation and whatnot. Now, it has given you this code over here. Let us see. Let's keep going, right? Uh you just transform a pipeline, global model reference, blah blah. Looks really good. Okay, so you can look at here. Let's say this. All right, run it. Load model once. Uh start up fast API. So, it uses Transformers pipeline to kind of inference this model. Looks good. It depends on what kind of model we are also using. It also gives you the deployment guide, so we can go through, like let's say, we can containerize that model through Docker, and we can use that. That's also there. And this is how you can serve it using uh you can see Gunicorn. I don't think people are using it nowadays. There are better inference servers like Triton and whatnot to use it. So, you can use Triton servers to kind of do that for inferencing, but this is what it is. And same way, you can do it through DashScope. So, I'll show you that in a

bit, right? The first thing that you have to do, you have to come to this Alibaba Cloud Model Studio. Like this has to be your landing page, you know, in case if you want to kind of create an account and get the API key. So, they say claim 1 million free tokens per model. So, of course, that's their elevator pitch. Now, this you can get it from here. Now, you start free, so you can just log in over here and start getting it, right? Uh so, just click on start free, and it takes you there here. Okay, this Model Studio. So, I'm already logged in, so I already have an uh you know, an account over here. You can see this is how it looks like, right? Uh and this is you get your API key and everything that you can get it. Okay. Now, in the playground also, you can test. Okay, you can get your API key from here. You can see. Okay. So, I have to log in now because it logged out because I had logged in quite a bit uh ago. Now, first you go there and get your API key from this uh Cloud Model Studio. So, I'll give the link in description. So, this is the latest model. You can see native multimodality. It says a leap in coding capabilities. Like for coding, it's fantastic, right? That's what they have been saying, right? So, this is where you have to get it. Now, you can also go and start using it on, like, Qwen Chat. Like, you know, they have a chatbot where you can try it out multiple things. You can see that's the default model for now, Qwen 3. 6 plus. That's the latest large language model by Alibaba that you see over here. And the same thing you can also do it. I have been trying it for a while to test it out on different, you know, queries and whatnot, right? This is fantastic. It also has visual capabilities, right? So, you can give images and then everything, and it can analyze it. Now, I am using through OpenRouter, so that's the interesting thing, right? So, OpenRouter is giving it for free. So, if you look at here, this is the link. It says Qwen 3. 6 plus free. So, it's free right now, right? Released on April 2nd. And look at the context window, which is fantastic, right? And right now, we do not have to pay anything. It's like completely free. Uh if you look at the ranking on OpenRouter itself, ranked number one at the in academia category, finance one, health number six, legal, and so on and so forth. This is where you have to use it. Now, I'll show you a code how I have built this, right? So, as I said that I'm running the backend in fast API, so I built a fast API endpoints, and then we have this simple React interface to kind of integrate the backend APIs uh into this interface. So, if you come over here, right? We have a backend folder, and we have a frontend folder. So, we inside the backend folder, you know, we have couple of files. We have a main. py, and we have requirements. txt. So, in requirements. txt, these are the dependencies we need to install. As we are using fast API, we need to use fast API. We are using Uvicorn, so Uvicorn and those are all fine. You just install it. Now, I have my API keys of both OpenRouter and the DashScope. So, DashScope is nothing but the Alibaba's Cloud Model Studio's API key that you use it and to use it to in, you know, start interacting with all the models made available by Alibaba, right? From that's from their server. So, if you look at here on OpenRouter, I have that, and same goes, you know, with DashScope. And then we have some middleware just to handle the course. And this is the model name. So, make sure that you are using this model name, you know, properly, because OpenRouter has like this, and DashScope has like this. Let's say we just do that. Right? So, we have our chat request just to handle the data structure through Pydantic. Uh we have this API key and all just to make sure this is, right? Uh fine. Now, with DashScope also, you have to see which region you are. So, if you're in Singapore region, if you're in Asia, if you're in Frankfurt, you know, this uh base URL might change. So, please go through the documentation. I of course I'll give the link in description, right? And you can just fork this GitHub repo and start building on top of this as well, right? And then we have a post request that we're just making to kind of use that. This is how we use it if you look at over here. So, OpenRouter is like this, and then DashScope is like this, right? And that's what it is, guys, right? We have a streaming protocol. We're using SSC for streaming protocols. And that's what it is. Very simple main. py. I'm not going to show you the frontend, which is probably not worth, you know, you don't probably care about this interface. The same goes with DashScope and OpenRouter, right? You can start using either of these two you want to use it, right? So, first go ahead and, you know, create your account. And you can try it out on the Qwen Chat as well, the same thing that I was talking about, right? Uh you know, you can upload an image, and you can start working out, right? On this. So, you can play with this uh model here as well, right? If you want. This looks really good to me. I'm very happy, you know, I I'm Qwen's model are one of my favorite models. And whenever Alibaba releases a model on Qwen, I always try it out and see how it's working, right? So, I'll just give the link in description. Please go ahead and check it out and let me know what you think about this new

model. I am very fascinated about Q and small even the open source ones, right? Which you can also of course quantize it further or fine tune it through unsloth library and what not. So, I'm always a big fan of Q and to be honest, right? Let me know how you are using this model and what's your feedback of this model, right? If you like the video, please hit the like icon. If you have not subscribed the channel yet, please do subscribe the channel, guys. That motivates me to create more such videos in the near future. Find all the links related to this model in the video description. That's all for this video. Thank you so much for watching. See you in the next one.