Googles NEW AI Gemini 2.0 Details Were Just LEAKED!

So the Verge has actually come out with another article in which they discuss leaks pertaining to Google's next Frontier model. Now this video I think is rather fascinating because when we look at this article, it actually talks about some things that we kind of didn't expect from the next iteration of model cycles. And I think things are about to change, but not in a way that most people do think. So you can see here that it says Google plans to announce its next Gemini model soon. This is of course something that we knew was on the horizon given the amount of time that it has been since the last model release. And it says that December is shaping up to be a month of dueling announcements from OpenAI and Google. And of course it talks about how Anthropic wants to take over your computer and Satya Nadella's good year. But the main focus of this video is going to be Google because of course, as you know, this time last year they released Gemini One Point Pro, which was outstanding. It managed to surpass GPT4O. There were just a bunch of things that were just absolutely incredible. But now all eyes are on Google again to see what do they have for their next Frontier model. Now, there's a lot of speculation going on, but I've been doing some research, some digging. I found a few things and I want to share them with you today. So let's dive into exactly what could potentially be coming in December for Google and of course, the next iteration of Frontier model. It says here that as my colleagues Kylie Robinson and Tom Warren reported, OpenAI is eyeing a December debut for its next flagship model. And fittingly, Google is aiming to release its next major Gemini 2.0 model in the same month. The source is familiar with the plan. Tell me. Of course this is going to be the continued trend as Google and OpenAI continue to try and one up each other. If you aren't familiar, you have to remember that this time last year it was Google that managed to release Google Gemini Pro, which actually surprisingly demolished OpenAI's GPT4 at the time across several different benchmarks. This was as to many people's surprise, as it demonstrated advanced reasoning capabilities across coding, across different contexts and across long context solutions. Now, with that being said, it seems like not only is OpenAI and Google doing this, but anthropic and X AI and of course Meta are racing to debut their next Frontier models. Now this is pretty interesting considering the fact that we've actually still been getting a frequent amount of updates with OpenAI and with Google with these Frontier companies. We've recently had Metaslama 3.2, OpenAI's 01 and even some secret Google models that seemed to slip under the radar for most people. So I will be talking about that later on in the video, but I want to show you guys a part of this article that might surprise you. Now do remember though that this is pure speculation, meaning that these articles could have the information completely wrong. That sometimes does happen and it's the only article that is reporting on this. But they are saying unfortunately that the next iteration of models aren't showing the performance gains that that these teams are hoping for. You can see it says that apparently OpenAI CEO Sam Altman is doing a phased rollout of the successor to GPT4, starting first with their business partners. But the sources apparently say that Google is planning to just simply widely release the next version of Gemini at the outset. Basically meaning that there's not going to be like any pre access, there's not going to be any special accounts, it's just going to be a blanket release, meaning that everyone gets access to the new model at the same time. Now if these models aren't going to provide the reasoning capabilities that many have suspected, it would mark a very interesting turning point for the AI industry. And I think this is going to be one of those ones that we're going to have to wait and see. The reason I say this is because there's often so much speculation about models before they come through and oftentimes sometimes they are accurate and sometimes they are completely off. Now I do think that if the model isn't as good as we think, I think that of course once again they may iterate to the next paradigm, which is of course test time compute or inference time compute, where they actually use the compute at inference time to get better reasoning abilities. After all, Google did publish a paper basically stating that this is the direction that they might be heading in. Now you can see that this article also states the fact that this chatter is the same that is happening across companies developing leading large models. Which means that once again it seems that if companies are starting to pivot away from traditional LLM models and move towards inference or test time compute, then it does mean that OpenAI would be leading the charge once again with as to where the next iteration of models are going to go. Now there was a screenshot from a small event where I did see Logan Kilpatrick, the lead product for Google AI Studio, who's working on Google Gemini's API and AGI, showing us exactly where we are headed. You can see that he has on the presentation Google's Gemini 2.0 saying that these next generations of frontier models are to be decided. They've also got the ability to tune everything and of course Gemini 2.0, Veo Imagen 3 and various things like NotebookLM. So it was clear that of course the next iteration of models is going to be Gemini 2.0. So at least we know the name of the model and what we can kind of expect. Now the thing is though, even if these models aren't as crazy in terms of their reasonability, I think Google is remarkably underrated in their ability to provide you with a variety of different things. For example, just because they might not be as useful. I just think Google has a UI problem because if you actually use the Google AI studio, you can input a variety of different things, things audio, video, image, and you can put things like videos that are an hour long, audios that are hours long. This is something that we simply just don't have from any major other platform. So I completely wouldn't bet against Google in that regard as they still do have some serious innovations under their belt. Now there was also a very cryptic tweet from Jimmy Apples which was rather fascinating. That actually does talk about Google's model. And if you aren't familiar with why I'm referencing Jimmy Apples, it's because this is someone who last week knew that Anthropic were going to release something API related and agentic. He says that someone may have got too drunk and said Gemini 2.0 has already been deployed to select B2B customers, but this might have been the ramblings of a drunk. Honestly, I find it sometimes hard to decrypt what he's saying here, but maybe Gemini 2.0 has already been rolled out to select B2B customers for initial testing. I mean, this wouldn't be the first time a major company has done such a thing. And of course Google might be trying to avoid any kind of backlash statement do get, as Google has faced numerous levels of scrutiny for having models that have been deemed a little bit too woke in certain scenarios. Now, that's not my biggest concern. My biggest concern is whether or not these models are going to live up to the height and actually become a part of people's everyday use in the AI ecosystem. Now, like I said before, even if these models do not live up to the quote unquote hype, I still think that Google are going to win this AI race in many different areas. For example, I'm Also hearing that Noam, the famed AI researcher that Google paid an ungodly amount of money to abandon character AI, is working on a separate reasoning model to advance OpenAI's 01. However, Google declined to comment on this story. Now, I'm not sure if Noam is going to be working on this or not, but I would say that it is most likely that Google is going to be working on an advanced AI model because they aren't just going to be doing one aspect of AI at all. So Google have brought us the innovations that have been long context and they are continuing to bring us a number of other breakthroughs that are going to make these AI systems more useful. The only thing is that when we're going to get these in products, I don't know. There are a few things that Google announced at recent events that some people may have forgotten. And I'm going to use this time to jog your memory on a few things that Google have in the pipeline because everyone seems to forget. Google are a terrific company that seriously has a lot that we really shouldn't bet against.

Okay, let's do some tests. Tell me when you see something that makes sound.

I see a speaker which makes sound.

What is that part of the speaker called?

That is the tweeter. It produces high frequency sounds.

Give me a creative alliteration about these.

Creative crayons. Color cheerfully. They certainly craft colorful creations.

What does that part of the code do?

This code defines encryption and decryption functions. It seems to use AESCBC encryption to encode and decode data based on a key and an initialization vector. Iv?

That's right. What neighborhood do you think I'm in?

This appears to be the King's Cross area of London. It is known for its railway station and transportation connections.

Do you remember where you saw my glasses?

Yes, I do. Your glasses were on the desk near a red apple.

What can I add here to make this system faster?

Adding a cache between the server and database could improve speed.

What does this remind you of?

Schrodinger's cat.

All right, give me a band name for this duo.

Golden Stripes.

Nice. Thanks, Gemini.

And one of the things I do want to leave you guys with is the fact that if you haven't been using Google, Gemini, maybe you might want to reconsider that one thing that did slip under the radar. And at the time I didn't even make a video on this because I didn't even realize it happened until a little bit later. And it was something that just missed the mark for me in terms of news that I was aware of. So Google did release an updated version of Google Gemini Pro that did actually perform significantly better on a range of different benchmarks, including math, including natural code. You can see that the mmmu, the Math Vista, even on benchmarks that other models can't even compete in, the models do perform exceptionally well. Just to be clear, this is Google's Gemini 1.5 Pro 002. So if you go into the Google AI Studio and you want to take a look at Google's more advanced model, the one that is actually quite useful, I'd say you go ahead and go to the AI studio because it's a place where you definitely want to be to get the most out of Google's products and services. So let me know what you think about this stuff. Do you think that Google still has it in the bag? I still am not writing them off at all. After all, they've got many different products in the pipeline and the recent debut of NotebookLM has left many people surprised and of course satisfied. So that being said, let me know your thoughts comment down below. I'll see you in the next one.