Speaker 1: Hello everybody, welcome back to another AI video. This one we're in chat LLM teams and today I want to talk to you about DeepSeek R1. You guessed it, chat LLM has already integrated it within hours of its release. Now, for those of you that don't know what DeepSeek R1 is, it is quite literally the best open source AI model on the market. It just launched like a day ago and it is lighting up the leaderboards, it's lighting up Twitter or X.com, whatever you want to call it. And it has so much going for it. It is 10 to 20 times cheaper to run than the O1 model, even though it is nearly as good. It comes with an amazing technical report. It has an open source license with an MIT license actually. And here's the other thing, they didn't hype it up, they just shipped it. There was no vague posting on Twitter, there was no strawberries, no nonsense. Open source is here, they're putting the heat on guys and I'm going to show it to you. Let's dig in, let's take a look. Alright, so to get to DeepSeek R1. Easy to do, you just got to log in to Chat LLM Teams. Easy stuff. Head over to their website, abacus.ai, links in the description below of course. And then just click on Chat LLM right here. Once you do that, you can either sign in with Google or your email or if you're logged in like I am already, you'll get this front screen here. And then watch this, I'm going to click this little down arrow up at the top here. And here we go, DeepSeek R1. It's literally that easy, it's integrated, it's ready to go. Now, let's look at some prompts and have some fun with it. Alright, we're logged in. Time to cook. Now, like I showed you quickly, I like to select the model first here. When I click on that little arrow here, I can select any model I want. Right now, I've got DeepSeek R1. But a lot of the times, I will select O1 or CloudSonic 3.5 and then do a comparison. So I'll run it in O1 and then I'll run it in DeepSeek for example. And then see if DeepSeek really is as good as they say it is. And by the way, it is. But you can always do those comparisons. But there we go. So I'm going to go ahead and just basically paste in a prompt here. I've got one here. And here it is. Marble is put in a glass cup. The glass is then turned upside down and put on a table. Then the glass is picked up and put in a microwave. Where's the marble? Explain your reasoning step by step. That is literally all you need to do. I hit enter. And watch this. DeepSeek R1 is working. It's researching actually. How cool is that, eh? And then now, it's going to go ahead and cook up its answer. It takes a few seconds. It's not as fast as 4.0 obviously. But it's a lot faster than O1. So it's kind of in the middle there. And here it is. Step. By. Step. Plan. Which is nice. Because that's what we asked for. Analyzing the marble's position when the glass is inverted. Okay. Determine the marble's location after inversion on the table. Since the open cup lacks a mechanism to trap it. Okay. Confirm the marble remains on the table when the glass is moved to the microwave. All right. Check. And then here we go. It analyzes. It goes through it step by step. It gives you its reasoning or reasoning traces in there. This is absolutely awesome. And that is just one. Example. Let's take a look at another. All right. Did you guys see how fast and easy that was? That was amazing. And if you're an American AI lab, you are probably sweating pretty hard right now. Because, yeah, that's open source. That is amazing. You're going to want to read that paper and copy what they did because they're coming for you. Open source all the way, everybody. But anyways, here we go. Let's do another attempt here. I'm going to go ahead and copy another prompt. A nice simple one. Give me 10 sentences that end with the word apple. Apple. Okay. We've gone ahead and made sure we've got deep seek selected. Here we go. Let's click on that button and off we go. Now, this is a little bit of a simpler prompt, so this should be a little bit quicker. Let's just wait. I'll wait a few seconds here. I'll speed this up in post. I didn't even have to speed it up. Here we go. So here are 10 sentences that end with the word apple or have apple at the ending. And you'll notice here that of these 10, nine of them are correct. But number nine here is not correct. So the model isn't perfect. It's really, really good. And apple is the second from the last word. But again, it's really, really, really good. And that compares probably quite well to open a eyes. Oh, one. In fact, let's go ahead and take a look. This is what it's for, guys. Let's go ahead. Give me 10 sentences and with the word apple. Now, I'm going to run it here and oh, one. This is just an additional thing I want to show you. This is going to take a little bit longer because oh, one's a little slower. I'm going to come back when it's done and show you what they got. All right. Welcome back. And look at that. They got 10 out of 10. So anecdotally, yeah, oh, one is still the big dog. All right. Fair enough. And this is just one example. I highly recommend that you try out your own. But as you can see here for a fraction of the price, the open source MIT license, no fanfare. Amazing model here, guys. This is how you do it. I recommend you try Chan LLM teams. Log in, drop down to deep sea car one. Do some comparisons with oh, one try deep sea car one. Do that. Let me know what you think. Leave a comment below. Thanks for watching.