NEW OpenAI o3-mini is Absolutely INSANE…🤯

breaking news chat GPT and open AI have just released open ai03 mini it's now available in chat GPT and the API you can see it was just announced 1 hour ago and pro users will have unlimited access to O3 mini and plus and team users will have triple the rate limits free users can try O3 Min chat GPT by selecting the reasoning model and you can see an example of it right here one of the most insane things about this is the most powerful but most dangerous model open AI have ever released you can see here that before deployment they carefully assess the safety of O3 mini and we'll be testing it out today so let's run through the details the benchmarks Etc

so open AI 03 mini is the newest most cost efficient model in the reason series and this also includes fun and interestingly they've said that this includes developer features including function calling structured outputs and developer messages along with support streaming you can see the benchmarks right here so if we compare for example 01 preview versus 01 versus 03 mini high and medium right so 03 mini high is outperforming all the benchmarks outperforming 01 by quite a long way from 87. 3 versus 83. 3 in terms of PhD level science questions O3 mini high is outperforming all the models like you can see and for coding so you can see here competition coding opening I3 mini achieves progressively higher ELO score with increased reasoning effort all outperforming O3 mini and you can see the performance right here software engineering also and also for human preference evaluations so they've said evaluations by external expert testers also show that open ai3 mini produces more accurate and clearer answers with stronger reasoning abilities than open ai1 mini especially for stem you can also see how it performs on safety metrics right here so if we go directly

into chat GPT now we can see O3 mini and 03 mini high right so O3 mini is faster Advanced reasoning and O3 mini high is great at coding and logic so let's compare the two different options here so we're going to test them out right now I've said to chat gpt3 mini create a onepage website for The Niche blah blah and it has come back so quickly look at that thought about it for 4 seconds and then it came back and did the HTML which is absolutely amazing I've never seen a new model from cat GPT come out that quickly chat gpc3 mini High which is a more powerful model took 17 seconds to develop but we'll compare the outputs to see which one is better side by side this has blown me away by how fast and easy that was to use so we're going to grab the HDML now we can't preview it directly inside the chat so it doesn't seem like for example you can use a canvas directly inside chat chpt with 03 mini you have to test the HDML s but let's test this out see what we've got we have a website built in literally 4 seconds that's how long it took from over3 minute amazing looks pretty good it's good to go it's got links to our funnels that is working perfectly really didn't take time at and you can see how quick and easy that was used absolutely amazing Let's test this one out so that was the output from o03 mini now we're going to compare the output versus three High very similar outputs right there obviously it took a lot longer and thing to note here as well is you probably don't need 03 High most of the time right 03 is going to do his magic quicker and unless you're really testing the limits of the model then 03 mini is probably going to do its job right so this is 03 mini medium here's the other thing as well so you

can actually search a web now which you previously couldn't do with 01 when I tested out the other day unless you're using the 01 Mini model so if we say for example okay what are the latest AI headlights today let's test out the search feature doesn't seem to come back to me on the first attempt let's see if the second attempt works better something to Bear him not I'm just being completely transparent with you right the model is not going to be perfect day one and this is a lot slower as well to pull in the results on search just something to note there as well so we're going to compare both now it's grabbing the latest news now this is lagging behind a little bit so you can see inside the search here that's pulling results from December the 24th 2024 which is of course that's over a month ago so it's not the latest headlines from today so there are still problems that you can see inside chat ppt3 mini it may be at the level of PhD science but at the same time it's still getting a bit confused inside the reasoning but let's see what it comes back with it doesn't seem to have given me analysis it seems to be a lot slower on the search feature there let's see what we got yes it's given us headlines from 2024 even though we asked for the latest headlines today so you can see that the model is still not perfect right at the same time though also you can see it mentioned open ai's O3 mini launch inside the logic and reasoning right so it's doing the reason is do the thinking there but there is no mention of that inside the latest headlines from today which is interesting so I thought why wouldn't it pull in the latest details so it's okay but honestly I think you're probably going to get better results with perplexity it would be interesting to combine the power of reasoning with today's search results but you don't seem to get that inside O3 mini let's see what chat G did here we go this is much better right so it says blow are several of the most talked about AI headlines Making Waves as of today February the 21st so already the results are better it talks about quen 2. 5 and alib Baba's release March better so 03 mini high you can see a huge difference versus 03 mini let's keep going now I'm going to say to both

create the world's best Space Invaders game we're going to remove search we don't need search anymore also just one thing to note you don't get to attach files inside O3 mini so you can't insert files Etc I'm going to switch off search here as well and it's just so quick I've just never seen a chat GPT release that responds so fast that is one of the things that's blown me away today it's a lot faster than for example like deep c car 1 or any other reasoning model that I've seen O3 mini high is obviously going to take a bit more time because it has to think more but even then it's already finished by the time I finished talking so let's compare these results we've got 03 mini up here we have the Space Invaders game from O3 mini that we've plugged in and that does not work at all and those images are not working at all let's see if the actual game itself works it does but the background is very weird look at that background it's requesting images from imer which is no longer available seems like hallucinated code there or something like that so let's compare now O3 mini High we'll plug that in look at that now that sort of works but actually I'm not controlling the green one I'm controlling the white one which doesn't make sense and you can see it's a bit laggy doesn't seem to quite work right there so there's still bugs inside the code that I think honestly you could get better results from other models as an example let's compare that versus deep seek R1 so we've used same prompt it's taking us 12 seconds to generate that and then we've got the HTML code ready to go and if we run the HTML which we can do inside the chat you can see here it is working perfectly look at that now be mind this is a code from O3 mini high and it doesn't work at all right super buggy I can't control the Green Man whereas this one works perfectly as you can see so in terms of coding we're still getting better outputs from Deep C car1 so far in the tests however I do love the speed it was much faster to get a response from chat gp03 mini than it was from Deep C G1 let's keep going now so

what I'm going to say now is generate some generic AI fluff content and you can see this content right here the reason that we're doing that is because we want to get something that's 100% AI detectable if we run that through an AI detector like C gbt hit detect X that is 100% AI generated now what we're going to do is we're going to try and humanize that using this test right here so let's start a new chat we'll take the content the AI 100% AI generated content right there and when we're going to use this prompt which is humaniz this so bypasses AI detected 100% at the time must be 100% non AI detectable we'll plug that in and we'll do the same inside O3 mini High here we go I do the fact that you can use both models side by side that is great because you are going to save a bit of time it does make things easier and it's also really easy to compare the models so we have the content back now from O3 mini let's grab this copy in there copy it plug it into zero GPT hit detect text and ni still coming out at 100% AI generated look at that not ideal my friends not ideal chat gpt3 mini high is still having to think about this it's been 44 seconds so far so let's see what we get back we're going to use the same prompt inside deep SE 1 just to compare them see what we can do so we put exactly the same prompt inside deep sear 1 and we have the content back from 03 mini height so let's grab that we'll go into zero gbt detect text and that's coming out of 78% AI generated which is I thought he would do much better than this I've tested this exact same PRT inside 01 and it's done for far better before let me show you an example so if we go inside 01 we'll plug the exact same prompt like you can see and we also got the content inside deep c car 1 so let's see how performs versus 03 mini and 03 mini High bear in mind 03 mini 100% detectable 03 mini High 78% detectable that you can see let's see how deep see1 performs we'll grab the content we'll plug it in look at that 5. 3% non AI detectable deep seek R1 the free model and the older model is still performing better than O3 mini high in most of the tests I've done in now 01 actually refused the request and 01 mini doesn't seem to be on the list anymore so we can't use that has shocked me

all right so one final test let's see what it's like for writing obviously these are not models designed for writing but we've tested out for coding building games we've tested out for logic and humanization so let's test out now we're going to say inside this model create an SEO optimize article for this best SEO speaker for Content creation do this and I'll give you some Source context about me and then just like some writing Style Guidelines like for example first person right Alex MOSI Etc we'll plug that into O3 mini we'll do the same inside O3 mini high and we might as well plug it into deep SEC car1 whilst we're here so let's see what we got now so both articles have used the source context for me the Deep C car1 article is over here and the chat gp03 mini if we compare them so for example the title best SEO speaker real talk on becoming a standout in the field I wouldn't say that's as good is what makes Julian gold the best SEO speaker right now and why you should care so I'd say the titles are better here let's compare these so it says what are so chat GPT fre mini if we have a look at this section of content it says what are the real challenges you face are you tired of the same Jon do you feel overwhelmed by the buzzwords and endless advice on SEO blah I just don't think that feels as humanized as this one that says I built a s fig link building agency from my kitchen table now I've got a team of 50 two bestselling books and over 70k YouTube subscribers this content just feels so much more humanized than this one over here let's see what we got back from o mini High see if it's better obviously it's not a rating model like I said but we can test it out so ever I would say the content may be even worse in this one so it say says ever sat down with a cup of coffee and wondered who truly stands out as the best SEO speaker I get it I've been there in myself I do like coffee but it's just not as good it's not as good as the content that you've got from here which feels way more humanized so for example but Julian why should I trust you fair question let's topics I crush on stage right it's just so the content here inside deep c one is much more humanized and I would use this article also bear in mind for this article which I am ranking number one for the search intent is looking for the best SEO speaker right so deep seek R1 actually answers the search intent by recommending me as the best SEO speaker if we go inside the search intent of the content from O3 mini High it just starts talking about content it starts talking about like SEO strategies in the middle of the article which not really relevant to the search intent right then it starts talking about like practical tips for becoming the best SEO speaker whereas what people are really looking for inside the articles is the best right answer which is what we've got for example Inside My article that ranks number one for this keyword so honestly I love the speed from O3 mini like that's absolutely great I'm just not that impressed by the results from my test so far but let's see maybe it'll be tweaked or find tuned I'm excited to experiment with the API as well so

thanks so much for watching if you want to get free access to all of my best prompts including 200 free chat GPT promps 53 aiso tools and a free course with over 230 video tutorials and saps on exactly how to use AI fors feel free to get that link in the comments description you also join a community of 3,700 members who are all interested in Ai and SEO like you see right here if you want to get a free one to1 SEO strategy session feel free to put there and we'll show you how we take websites from 0 to 145,000 business month and generate hundreds of thousands of dollars in sales and autopilot on this free link booing acceleration session you get a free SEO domination plan discover the secrets SEO link booing or answer any questions you have you learn the best link building strategy for your website plus how to c out ranking link building and how to turno traffic based on what's working for us make sure you check out the AI profit boardroom it shows you all my best Cutting Edge automation systems with AI that help you make more money and save hundreds of hours with AI Link in the comments description you can join now and it also comes with weekly coaching calls directly with me specifically focus on helping you make more profit with AI feel free to get that link in the comments description appreciate watching byby