# OpenAI just made some MASSIVE announcements

Channel: Alex Finn
Video: https://www.youtube.com/watch?v=cro5EDMlcTY
Duration: 21 min
Language: English
Words: 4245
Transcript page: https://viewrankai.com/tools/youtube-transcript/cro5EDMlcTY

---

[0:00] I just got back from OpenAI Dev Day. It was incredible. So many amazing announcements. I've been using a lot of those amazing announcements for a few weeks now. I was very lucky to have early access to a lot of them. I'm going to cover all the news. I'm going to cover everything announced, give you my initial reaction, tell you what it's been like using DOTS for a while now, tell you what it's been like using space, different things like that, and show you everything you need to get excited about and how to take advantage of all this amazing technology and boost your productivity and all that. It was

[0:33] awesome. If you ever get the chance to go to Dev Day, you should do it. It's a very, very fun experience. Let's just jump straight into it. Let's go into all the announcements starting with the biggest one which is Open AAI Dots. Dots is Open AAI's competitor to Grockbot and now more recently Meta Muses their own

[0:51] spin on it and I actually like it a lot. I've been using it for a little bit now. I was lucky enough to have early access and I got a ton done with it. You can see right here. Here's my Bobby. It is a great tool. What is DOT? Well, it is their AI agent, right? is your front door to chat GPT. Before when using chat GPT, you can see over on the left here, you create threads for everything. You want to respond an email, you create a thread. You want to build a game, you create a thread. You want to merge a PR, you create another thread. What that

[1:21] leads to is a lot of thread management. You end up with thousands of threads if you do a lot of work and sometimes it's very hard to manage. Now you have dots. This is your Metamuse Grobot. That is your front door to everything in chat GPT. Now, instead of managing a bunch of threads, you have your agent, your dot do the management for you. I need to send an email. I don't create a new thread to send an email. I go to my dot

[1:46] and say, "Hey, respond to that email." Then, at the same time, I'm building a game and I think of a new feature. I then follow up with that same dot, and say, "Hey, build a feature in this." And because it has context around all your threads, because it has context on all your plugins, memories, data, and all that, it's able to manage all of it for you. This, in my opinion, makes the chat GPT experience a lot nicer because instead of having to manage all these ideas and all these things, having to remember where the threads are, pin this, which projects that in this that projects, threads, you don't have to

[2:17] think about it anymore. It's all managed through your dot. So, if you are a chat GBT power user, this is excellent. This is great for you. But what's the difference between dots and the more popular bots right now, uh, Metamuse as well as Grockbot? gonna be honest with you, not much. So then why would you switch? Why would you put all that

[2:37] effort to switch your agents to dots? Well, there's a few reasons why. Number one is memories. Your dot right out of the box, right when you start it, has all of your chat GPT memories. So if you're anything like me, you've been using chat GPT for years. You have tons of conversations, all that context, all those conversations. Your dot knows it right away. If you started using MetaMuse over the last few weeks, it knows literally zilch about you. You had to tell it a ton about yourself and still probably doesn't have your entire context. Dot, on the other hand, has everything right out of the box. You

[3:10] don't there's no onboarding. There's nothing you need to do. It just knows everything about you. That's pretty big. The second moat it has over other bots is the fact that it's in the best agent harness on planet Earth, which is the Chat GBT desktop app. The Chat GBT desktop app is incredible. It has the best tools AI agents can have. Between the voice mode, between all the plugins they have already, between something I'm about to show you in a second, Chad GBT

[3:36] space, it is the best agent harness. It's the best to use. It's the best to get work done. Your DOT has access to all of those tools. So, you can go to your Dot, say, "Hey, spin up a codeex thread for this. Spin up a thread for this. Do this in the cloud. Do this on

[3:50] my local computer. Do this on my iPad. Do this on my cell phone." Whatever you want. It goes across all those. It can do whatever it needs in all those. It has the best tool set of any AI agent out there. It can work on the cloud very much like Metamuse and Grobbot can do. But it also can work locally. Any computer that has Chad GBT on any device, iPhone, anything your agent, your dot has access to. So that is a big

[4:16] moat and a big differentiator as well. Functionally, feature-wise, is it any different than anything else out there? No, honestly not. I mean, it's just basically Metamuse, but with Chad GBT memories as well as all the tools from Codeex. But I still really have enjoyed using it. Who is this for? Who should switch to DOT? Well, if you're a Chad GBT power user, you should 100% switch over. It is fantastic. or if you're any sort of AI power user at all, I think you should switch as well because of all its power user tools. I think Metam Muse is more for the make me a reservation crowd out there, which is honestly

[4:55] probably a way bigger crowd than the people who use like Codeex, but still I think it's more for kind of the more casual AI user that uses AI to make reservations and buy things off Amazon. Grockbot, I still think it's a fantastic tool. If you like having multiple personas, then you probably should stick with Grockbot because it's only one dot inside Chad GBT, which might be a positive for you, it might be a negative for you. But if you like to have that huge team of AI agents, Grockbot is really good for that. Which brings us to the next big announcement I've been using for a little bit now, and that is

[5:26] chat GPT space. This is an interesting one. This is actually a major major swing by Open AI. What space is is basically artifacts on steroids. So every AI agent tool out there, every claw code, everything has what's called artifacts. You're chatting, it makes docs for you, makes slideshows, it makes games, whatever, and it shows up in your chat. That's called an artifact. But this takes it to the next level. Let me show you an example. So this is a space my dot Bobby made for me this morning. I woke up, it saw I was going to debate. I

[5:58] didn't ask for this. It just made it. And it made me this entire itinerary with all the times, everywhere I need to be, everything I need to do to get going. Pretty cool. And also a big feature of dot just this proactive stuff. Really, really cool. But here's where it goes to the next level over artifacts, over what you'd see in really any other AI tool. This is one interactive, right? So I can click in here in any of these artifacts. I'm going to start typing. I can say, hey, let's actually visualize my day. So I may I want to come in here and visualize my calendar for dev day. And I can hit

[6:29] enter. Now it's going to add a generated chart in my space of my entire calendar. Couldn't do this type of editing before in different artifacts. I can take it another level. If I want to just chat with this page, if I want to chat with this space, I can say I can say, "Hey, at the bottom, add in all the notes and announcements from dev day." I hit enter. Now I'm chatting back and forth with uh the space and adding to it, which is really cool. But taking this even further, and this is where it gets really, really interesting, especially if you have a team or anything like

[7:02] that, this is completely multiplayer. So, you can go in here. I'll minimize this. Also, for the record, look at that. That's sick. It put my entire calendar from actually uh my Google calendar in here and then added in the schedule from dev day. You saw that happen in real time. That's really sick. Anyway, it's multiplayer, so you can click share. You can share this to other people and they all can work on this with you. And not only that, you can actually tap in each other's dots to work on this too. So if I want to go in here, I want to say at Bobby and say add

[7:36] in the weather and I hit enter on that. Bobby, my dot will now go in and actually add in the weather to this space. So you're collaborative, right? It's fully multiplayer, which is kind of the big trend in AI now, making things multiplayer, so multiple people and agents can work together. And now you all can collaborate on these different documents and artifacts. They're coming after Google Docs. They're coming after notion. And they're putting this all in chat GPT. And they're superpowering it with AI, making it all multiplayer, which is incredible. I think what they're really trying to do is make it so all your work, everything you have to

[8:11] do on your computer, you can just do it straight in chat GPT. I think they're trying to find every workflow people do where they have to click out of chat GPT and having them just put it in here so you spend your entire day in chat GPT. That's really amazing. I really like this. Here's my concern. I mean, that's a tall task getting people out of notion, getting people out of Google Docs and all that. You're asking people to completely change their workflows, which for me, based on me using this, I think is worth it because it's so sick because of all the collaboration. But for a lot of people who are like stuck

[8:41] in those tools, who've been using those tools for years, it's going to be a tall task, but we'll see. I think this might stick. Every year, OpenAI makes like announces like 40 different things. 75% of them don't really stick, and by a month later, they're gone. I hope this is one of those things that sticks. I really like using this. Which brings us to the next announcement, which is a brand new model from OpenAI GPT 6.1 Soul. This is kind of mind-blowing. The biggest reason is GPT6 Soul dropped, I think, like 5 days ago. Now, I don't think there's ever been this quick of a turnaround time uh for a new model from

[9:20] really any company. It's obvious why they did this, right? The reason being is uh Opus 55 comes out, I think radically changes the entire industry. I think shocks basically every AI company out there uh with how big of a massive leap it is in basically every single way from intelligence to speed to cost. It dominates every other model in all three of those categories. Less than 24 hours after that comes out. Open AAI releases GPT six soul. It is a disappointment compared to Opus 55. I don't think OpenAI saw Opus 55 coming and they had to quickly scramble because for the first time in a while there was really

[10:01] no reason to be using GPT models and so they had to scramble and in the last 5 days I wouldn't be surprised if OpenAI employees did not do much sleeping the last week. They come out with GPT61 Soul. Now I'll be honest I was able to early test a lot of the things I'm showing you here in this video. This was not one of them. I did not get my hands on GBT61 soul, but the benchmarks are looking really nice. It's looking like it is basically Astra 6 level, if not even a little bit better for an absolute fraction of the price. Basically what Opus 55 was to Fable, Fable level

[10:38] intelligence, but significantly faster and significantly cheaper. This looks like that version for Astra. Basically what people are expecting out of six soul, let's look at what 6.1 soul is. Cost per task is absolutely dirt cheap and it's looking like the speeds are amazing. Now, how does it compare to Opus 55? Can't tell you. I need to do some testing. I will probably release a

[10:59] video on this in the next couple days. So, make sure to subscribe and turn on notifications. I'm going to do dedicated videos to every single thing and announcement I'm showing you in this video over the next few days. I kind of want to just show you and go over high level the announcements also kind of flex that I went to dev day. Not a big deal. And then we will do the dedicated super in-depth deep dive over the next few days. Which brings us to the next big announcement and this was a

[11:25] controversial one to say the least. There is a new plan uh in Chad GBT, a $500 plan. Now, this is especially controversial because earlier in the day, they made a not so warm announcement that they were cutting the usage of the Chai GBT $200 plan in half. Bingo, bango, cut in half. You get half the usage you normally get. So, they make that announcement, which quite frankly pisses everyone off. And then they come back and say, "Oh, by the way, you can now pay us more money uh for a little bit more." So, you get 25x the

[11:58] usage. The $200 plan previously was 20x. So, you're really only getting 20% more usage for like over double the price. The other big benefit here though is ultraast. So, you get ultraast, which is their new mode. You previously got fast mode, now you get ultraast mode. What is ultraast mode? Well, it's only available in six Astra. And if you go to six Astra and you click speed when you're in the $500 plan, you click ultraast, you now get 8x speeds for 6x the price. That's right. Now it's eight times faster than your normal inference. I tested it earlier. I built an entire game using ultra fast. I think it burned like 6% of

[12:40] my usage, my $500 plan usage, but it built it spectacularly fast. It built like a full in-depth Unity game in like 2 minutes. It was unbelievably fast. Now, the issue everyone's going to, you know, kind of mention in the comments, but $500 is absolutely absurd. No one can afford that. You're right. That is very, very expensive. This is for select people. This isn't for everyone. I will say this 6 Astra, it's a fast model as it is. You don't need the 8x speeds. I was blown away by the 8x speeds. You don't need it. But there you go. Very

[13:13] controversial announcement. $500 plan. You get ultra fast. You get 25x usage. If you are a power user, I think it's I think it's a mustave if you're like a super power user. I think you got to do it. There you go. Let me know down below. Is that crazy? You sign up for the $500 plan, you sticking with something cheaper, you just switching to Claude? I don't know. Let me know down below. Which brings us to our next announcement, which is a vastly improved codec. So, this is really cool. I'm glad they did this. I'm actually using the CLI more and more lately. Reason being

[13:44] is I use Herder, which you can see here. If you want a dedicated Herder video, by the way, let me know down below as well. It Herder basically allows you to manage a whole bunch of agents very, very easily all at the same time. But you have to do in the CLI. I use the codec a ton. They vastly improved it. It looks a lot nicer. And best of all, you can now do voice. They have voice mode in the CLI. For those who watch my videos, I talk about voice mode a ton. I think it

[14:09] is mind-blowing. I think it is amazing. It was only available in desktop app though. Now I can do it in the CLI, which is so so sick. So CLI visually improved and now you can do voice mode in it. I highly recommend checking out Herder. It's a free open source tool that let you just manage a ton of agents at the same time. I use it a ton and now I can use voice mode in it which is amazing. Another big addition to Codeex today was Codeex in the cloud. Now inside Codeex, you can spin up tasks and immediately send it to the cloud, which is great if you're one of those people

[14:40] that have your laptop open at all times just so you can get tasks done. All you need to do when you're spinning up a new thread in Chad GPT is you click the little thing here that says local and now you can switch to cloud. And what this will do is whatever task you give it, it will now immediately send it to the cloud, send it to server. So I can lock this computer, shut it down, doesn't matter. You can be assured your task will get done because it's happening somewhere else. That's really cool. Really nice little quality of life addition. Here is a big one that I think

[15:09] will be huge if you are an app builder, and that is the plug-in extensions. They are doubling down in on plugins which they should be doing and I think the the popularity of agents is kind of forcing this but it is the ability to build full experiences inside of Chad GBT for your application. It allows you to basically bring all the functionality app straight into Chad GBT where most people are doing their work. And this is big for distribution right there's billions of people that use chat GBT. So, if you have an app, you're building some sort of application, you can create a plug-in so that those billions of users can now

[15:44] get access to your app inside of chat GPT. Again, we talked about this earlier. It really seems like OpenAI is trying to make it so you never have to leave chat GPT. This is basically the app store of Chad GPT. This is actually something OpenAI's attempted many times before. They actually announced at the last dev day I was at a Chad GBT app store. It never panned out. People didn't really use it. So, they're trying again. They're giving it another shot. I think this is the winner, though. I

[16:13] think this is the one that catches on. Reason being is people are getting into plugins now between Muse, Dots, all of them. Everyone's trying to build connectors and plugins so that their agents can access it. This seems like the best version of it. This seems like the one agents will want to use and click around in. It's almost like a space for your app. I think we now live in a world where if you have an app, like I have my app Creator Buddy, I'm

[16:38] building out Henry intelligent machines. I think I'm going to need to build a plugin so people can get access to it in chat GPT. It just opens up your distribution. It makes it so way more people can get access to your app. There's going to be so much discovery that goes on with this plug-in store where people will be able to go in and

[16:55] discover your app and start using it. You got to be building your own plugins here if you have your own app. Here is one that was awesome. This is a really good announcement. just a proconsumer announcement which you don't always see a ton from these AI companies and that is sign in with chat GBT. What does that mean? You can now sign into other apps

[17:13] with chat GBT. But this isn't just OOTH. This actually allows you to use your chat GBT plan in other applications. So you remember when like OpenClaw dropped back in like February or January and Anthropic immediately went no hell no you can't use our subscriptions with that. You can't use it. we don't know what that tool is. That pissed a lot of people off. Well, OpenAI is going the opposite route. They're saying, not only can you use our tool with Open Claw and Hermes, you actually can use it with

[17:39] whatever the hell application you want. So, now you can take tools like Devon connect to with Chad GPT and now you can use your inference from your subscription plan in other tools. So, they want you using your subscription for absolutely everything. And this is pretty cool because this basically allows you to demo or trial other AI tools out there. Instead of paying for Devon's subscription plan, you just sign in with your tool and you can use your inference for Devon. If you decide you like it, you can buy their subscription, get more usage out of it. Very, very proconsumer. Allows you to use your subscription in more places. Use your

[18:15] inference in other tools or instead of having to get more and more subscriptions, you just use your one chat GBT subscription. That's really sick. It's taking the opposite approach of what Anthropic's been doing in the past. They're going proconsumer, which I love. Really amazing work by Open Air on this one. There were many other announcements here, including private intelligence. Basically, a way for you to talk to chat GPT without it storing anything, any of your data, anything you're talking about at all. There was also an adventure mode they talked about

[18:45] very briefly, which is really cool. Basically, a very easy way to build games, have the AI test your games. I think gaming is going to be a massive, massive, massive application AI. It's a huge industry that is ripe for disruption. I think a lot of these companies, Anthropic, OpenAI, will be coming out way more gaming tools. So,

[19:03] it's awesome to see them releasing that. I'm going to be sending out guides on all of this through my free newsletter. It's the number one AI newsletter on planet Earth. Link down below for that. It's called Ship It Weekly. Sign up for that news. Over 40,000 people subscribe that link down below. Subscribe and turn on notifications. Leave a like if you learned anything at all. By the way, they gave everyone this. This is so sick. This is a chromatic, a mod retro chromatic. I don't know if you're familiar, but it's basically like a Game Boy Color, but like like a superpowered one with like a backlit screen and

[19:32] everything, but what's cool is you can plug this into your computer and you can vibe code vibe code games on it. And it comes with this like OpenAI cartridge where you plug it in and it puts whatever games you build on the cartridge. So you can like vibe code like Game Boy games and play it on your handheld. 13-year-old me is going nuts over this. Uh, so this was really cool to shout out OpenAI for giving everyone one of these. Pretty amazing. I think Palmer Lucky made it. So really, really cool. Uh, if you ever get, again, you

[19:59] ever get the chance to go to Dev Day. Highly recommend it. It's a really cool experience. You meet amazing people. Could the food be a little bit better? Yeah, it could be a little bit better. Do I want to eat macaroni and cheese and salad for lunch? No, I don't want to eat macaroni and cheese and salad for lunch. But I can't complain. Everything else was amazing. Everyone got a free Mod Retro. You get to meet Sam Alman. you can do all these other cool things. So,

[20:19] really amazing. Hope this was helpful. Again, deep dives in all these features I talked about later this week. Wanted to keep you in the loop. It's an amazing time to be alive. It's an amazing time to be having fun with technology. I'll see you in the next video.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt