OpenAI Cancels o3

Craig: so we talk about models all the time. we were just talking about anthropix about to release their new model. yeah. xai just dropped their version three. so openai has decided that their normal o3 model, they're dropping it. they've got not o3 minutei or o3i, but then the regular o3. because of pressure from deep six, they're accelerating these features and rolling them into gpt four point five and gpt five. what do you think is going on here, john?

Jon: well, i think it's smart because the way that they're releasing things, it's confusing. three, they got four, all these different models. i think they're smart to focus on developing gpt four point five, gpt five, which is simplifying their product offerings. that makes sense. i think, and maybe this is why i've stopped using chatgpt as much, it's just i don't care to figure out which. i've got tons of other options and i think it's a smart decision for openai to do this. but yeah, the cancellation, it seems to be a strategic pivot for openai, which means that they're going to focus on integrating the advanced features into a more powerful system, which aligns with the trend in ai development regardless. so i think they tried to go this different route and it didn't work. what do you think though? because i'm curious what you think because you love using chatgpt.

Craig: well, it is my main go to from an ai perspective, no question. and the different models do different things and some are better at different things than others and some costs more than others. so it depends on what you're doing. like anything else, i think if you're using models for ai and automation, you pay attention to what each of them specifically does. if you're just using them for text generation, you go with the one you're happy with. when i say text, talk back and forth and that kind of stuff. if using one that's coding, like i said, you go with one that you're happy with that does the language and works with you. i think this is a direct result of what's happening with deepseek and other companies where they're starting to release cheaper, better models and they need to catch on with this. four. five is supposed to be coming out likely to release for the next few, it says months here, but i heard few weeks that we're going to get gpt four. five and it says potentially late twenty twenty five, early twenty six here for gpt-five, but i think it's closer to mid twenty twenty five, maybe late twenty twenty five for gpt-five.

Jon: i have to say, so we cover a lot about ai. i have to say, what is happening in the landscape right now truly shows what competition can do to help the consumer at the end of the day, that benefits the consumer. this is a direct example. other people say that i haven't really seen some really good examples of competition recently other than ai. you, but i think this is a great example of how competition is just benefiting the end user.

Craig: yeah, it's pushing everything faster. deepsea came out and shook the ai world. we were talking about a couple of weeks ago when it came out, all the stocks dropped and everyone's worried about it. then i haven't signed up. i've used it a few times, but i haven't signed up user api keys, but i will because it's a tenth the price, if not more than that, cheaper than openai on a budget than anything really. yeah. and so you sit back and say, okay, why wouldn't i use it for some things? and you can talk about that because it's china's just this and it's that. all depends what you're using it for, honestly. if i'm calling it to sort some stuff, if i'm calling it to do stuff, if the information or the data in there is just public information, then who cares? yeah. but other people will say you're promoting this, you're promoting that, but you go with what's best for you, the company and what you're doing. and it really changed the world because lot of these gemini dropped the new model, anthropics dropping a new model, everyone is dropping new models and their pricing is changing completely as a result of this. so yes, it is how competition, it changes and makes the market better. yeah. gpt four point five is expected to build on four o's functionality with performance enhancements in language understanding, reduced latency, more accurate context handling, refined ethical guidelines, enhanced transparency and ethical protocols will be embedded ensuring safer and more responsible ai interactions. not really sure what that means. i find ethical stuff usually gets in the way rather than helps. but that's probably just the way it's implemented. i say that because you ask something, say, create an image. openai does this to me all the time. i work on something. i say, i want some text for this article. text image prompts, want this article i just created. i'll say, here's the article, give me something. i'll fuss around with it. i don't care who owns the copyright like before. i'm just creating some social media images so i can post this stuff. it'll create prompts and i say, great. i like that one. like this one. i create image number one and i'll say, oh, i can't because it's got copyright. you know, like, why do you create image prompts that you can't actually generate? and it goes into some sort of weird doesn't really. and that's why you have to that. that's why i mean by the ethical things kind of get in the way because it generates the stuff and then tells you, you know what you're doing is wrong because it generated it. and i'm like, okay, well, fix those things and we can move forward rather than getting in the way. hopefully that gets aligned better. i like the idea that it's in there. i just don't like how it's implemented and to be fair, when it comes to image, don't like it. it's in there. i use grok mostly because it doesn't give a crap. because it'll create images with logos and stuff for what i'm doing. i'm pretty happy with that because it's trying to prevent people from being a company or doing something that's wrong or whatever. but here i am making youtube videos or making a podcast, and i want a couple of images that reflect that i'm talking about google or i'm talking about this, so i'll throw some logos or whatever in there. it's common practice is what goes on, but depending on how you use it, it thinks that's unethical. you want to use the thing that's right for you. better customization features that will allow users and enterprises to tailor the model for their specific tasks. oh geez, we're overtime already. sorry, john. robust integration, the model offer seamless integration capabilities with existing platforms and workflows, reducing operational friction for businesses. i'm not quite sure what that means because they're more and more talking about agents and i'm excited because building ai agents is great. and the more closer we can get to the actual ai model to do stuff rather than being removed from it, the easier things will get done. gpt-five revolutionary reasoning. gpt-five aims to exhibit advanced reasoning capabilities, improve logic and problem solving skills, surpass incurring models, creative flexibility, deep contextual understanding and integration of multimodal data. so expect to support not only text but images, video and possibly other forms of data making gt5 a versatile tool for modern applications. i'm excited for these things because it's what i use heavily. it's the one i pay for the most and i'm excited to see this happen.

Jon: yeah, me too.

Craig: again, whole thing is pushing forward and i really love to see it. love to see the whole industry moving forward. so let us know what do you think about this? is this a big deal to you? you don't use openai, you're using deepseek or one of these other models, you're using cloud log john and who gives a crap? or do you like to see these leapfrogs and competition happen because it drives this stuff forward? let us know your comments below.