AI News: Chat GPT 5.4 LEAKED, DeepSeek V4 Next Week?, #QuitGPT Movement!

Something crazy just happened over the weekend. Tibo, someone who works for the Codex team, was showcasing something in the new Codeex feature, which was supposed to be a secret, which is enabling an experimental feature for hotkey to pop up a mini window. That was initially his intention. However, what you can see in this picture is that he accidentally leaked GPT 5. 4 model. And what's crazy, if we go to that post where he originally posted it, that post does not exist anymore. So, how did this all happen? Well, this person works at OpenAI and he works for the Codeex team and he was showcasing a feature, but he might have just leaked a new model that we are waiting for. Currently, everyone has been waiting for the GPT 5. 3 model. It's been heard about. There was a leaked model name for it called Garlic, and everyone has been waiting for months now. But it might be because this new model might be super strong compared to the GPT 5. 3 upgrade and might be called GPT 5. 4 instead. What's funny is that this person over here, who's pretty wellknown in the AI news and rumor section, he said that it seems more and more obvious that the next OpenAI model will be GPT 5. 4 instead of GPT 5. 3. This is just pure speculation. And he actually posted the original post. And then if you go to the comments of that post, this was the post that he originally linked to. It was removed. All right. And that was the post I just showed you guys. But we can see that there is a lot of expectation that the new model is going to be much stronger than the old model is expected to be GPT 5. 3 and the new model being actually called GPT 5. 4. Thanks to this guy, Tibo. I hope this guy still has his job. But for us, that means that the new model should be coming out very soon. And when it does, it might be stronger than we might have originally thought it's going to be. So, I've been actually thinking about this and I just want to put this out there because I don't think enough people are talking about it. GPT 5. 3 Codeex just came out, right? And honestly, it's been pretty impressive. It's faster than the previous version, doing really well on coding benchmarks, and tons of developers are already using it. It's been a solid release for OpenAI. But here's what that kind of got me thinking. People and including myself have been speculating that the next open AI model is going to be called GPT 5. 3. And I get why people are landing there that sounds like natural next step from GPT 5. 2. But 5. 3 is technically already out there. That's the codeex model and that name is taken. So if we're just following the pattern and Open AI has been pretty consistent with how they've been numbering these things throughout the GPT5 family, the next one should just be 5. 4. That's it. Nothing crazy. just basic sequential numbering and it's probably very beneficial for them on the marketing stand as well since the 5. 3 name is already associated with the codeex model. GPD 5. 4 is probably going to be really interesting if this trajectory actually continues and we get a model that is not only better than a GPT 5. 2 model but also 5. 3 codeex model. I don't know, it just seems like an obvious thing that's getting overlooked in a lot of the conversation right now. We're probably already past 5. 3 and the next chapter might be 5. 4 and I think people are going to be surprised by how soon it arrives. Hopefully I'm right about this. Who knows? But I'm willing to take my risk on this bet. And this might be just

another pump fake news from China, but it seems that the Deep Seek model version 4 is finally supposed to be coming next week. Now, I know I've made many videos on this channel saying this week, this week, and I've been waiting for this model genuinely. Deep Seek literally changed the game when it dropped last year. It showed that you could compete with large Frontier models at a fraction of the cost. Now, this model was supposed to be actually launched during February during the Chinese New Year, but it has been, I think, been delayed a little bit, maybe because so many other Chinese models dropped during that time, and it might have made the competition for Deep Seek a little bit more challenging. Now, this model, I don't know if it's still going to benefit from a release at this point. Claude has dropped a new model. Open AAI is thinking of dropping a new model. Even we got Gemini 3. 1. So, we've got more Frontier models coming as well. More Chinese models that have dropped. So, this Deepseek version 4 definitely needs to change the game by a lot. The margin for success has gotten a little bit tighter for Deepseek because if they drop a model that is only competing with GPT 5. 2 or Cloud Opus 4. 5, then they're going to be left behind. They need to start competing with the new models. GPT 5. 3 that might be coming or GPT 5. 4 like I think and Opus 4. 6 now these models have taken the market share and Miniax and Kimmy all of these models from China have also taken the open-source market share. So, Deepseek has been quiet for a little bit too long, I think, and they might have lost a little bit of their thunder with the Deepseek version 4 release, but I could be completely wrong. And Deepseek version 4 comes out and as the leak mentioned, it's going to be stronger, going to compete with chat GBT, is Claude, but hopefully it's competing with the newest version of those model and it wasn't trained to compete with the older versions because all these frontier labs have been dropping very fast nowadays.

So, if you haven't heard already, Enthropic and the Department of War have been at a little bit of conflict internally over the last week, and I made a full video explaining this. So, if you haven't, make sure you check out this video where basically USA and Donald Trump have basically banned Enthropic from government use. Enthropic disagreed with the Department of War for allowing it to use the cloud models to create autonomous weapons as well for mass surveillance. But obviously over this weekend, I hope everyone is safe. But USA, Iran and all the war have been going on. And it seems like despite all the tension between anthropic and the department of war, claude was actually used during the Iran Iraq war. It was used for intelligent assessments, target identification, and simulating battle scenarios. Now, this is just the future of where AI technology is going to be used. You might not agree with the Department of War. Donald Trump. And you might agree with Enthropic or it could be vice versa. But the reality is AI is going to play a part in all aspects of our lives. It's going to take part in war either you like it or not. It's going to be used by all parties, good parties, bad parties, depending on what side of the equation you are on. But AI is a tool that is going to change a lot of things. And war is something that's also going to go under some change. Obviously, the general public has no idea how these AI models are actually going to be used. Well, from earlier assessment, it looks like it's going to be used for simulation, battle scenarios, maybe stress testing theories, things like that, as well intelligence, surveillance, and even maybe in the future autonomous weapons. All of these projects are obviously very classified and obviously all these companies have NDA agreements and the projects are not going to be public knowledge until maybe they're deployed. But we can see that the line between government and AI is just getting blurred and these technologies are going to be used by the government even if you don't like it. Or

after Enthropic was blacklisted, OpenAI and Sam Elman basically struck a deal with the Pentagon. Now, and this is the most important part, OpenAI says that their deal actually includes the same two protection anthropics was asking for no mass domestic surveillance, no autonomous weapons without human control. And they published the contract language. They're saying their agreement has more guard rails than any previous classified AI deployment. Alman even publicly said the Department of War should offer these same terms to Enthropic. So the question everyone is asking why did the government accept these terms from OpenAI but not Anthropic? Nobody has a clean answer to that yet. What we do know is that the public responded Claude shot to the top of the app store on Saturday with people publicly migrating away from Chat GPT. Celebrities including Katy Perry apparently weighed in. A protest is being organized outside OpenAI headquarters and Samman held a ask me anything on Twitter that got 5. 6 million views where he reflected on the debate around where governments or private companies should hold more power over AI and said personally he doesn't think that's a close call. Meanwhile, Anthropic statement was pretty unamiguous. No amount of intimidation or punishment from the Department of War will change our position. and they're planning to challenge the supply chain designation in court. This is a story that is very still much unfolding, but one thing is clear, the AI industry just got pulled directly into the center of US politics, and nothing is going back to normal after this week. But that's it

for today's video. Make sure to subscribe to the channel, follow us on Twitter, follow the world of AI, and don't forget to subscribe to our newsletter. We post constantly, and you don't want to miss this. I'll see you guys in the next