NEW OpenAI o3-mini Beats Deepseek R1!?🤖 FREE Super FAST Reasoning Model USA vs China AI Wars!

All right so we've been covering R1 a lot on this channel lately and frankly all the AI industry and really just the mainstream media ever YouTuber is creating content about R1 because it has completely disrupted the industry and China is really going hard with quen 2.5 Max and other models and this has essentially forced open ai's hand to drop open ai03 mini pushing the frontier of cost effective reasoning to the next level they just dropped this a few hours ago and

The benchmark s are impressive you can either use the low medium or high model and guess what they're rolling this out to free users as well now personally I don't think they would have done such a big push and roll it out to free users if it wasn't for deep seek of course and all the hype behind it we're going to dive into this model show you how some people are using it do some different tests and see how it compares

To deep seek R1 is it better or does open source still have the upper hand let's dive right into it all right guys so like I mentioned this just dropped a few hours ago where they actually released O3 mini so news is still coming out people are still finding use cases and I'll definitely probably do more videos surrounding this let's go over all the information that we can find right now about this model so all links I cover in today's video

Will be linked down below so I'll leave a link to the system card if you really want to go through read some of this and check out some of the nitty-gritty details now just as a side note like I said you can try o03 mini free of charge in chat GPT by selecting the reasoning feature in the chat bar and the rate limits will be similar to the existing GPT 40 limits so 03 mini will also be available for chat GPT

Plus team and pro users worldwide today and open AI is tripling message limits for plus and team users to 150 messages per day only Pro users who are willing to pay $200 a month will get unlimited access to 03 mini so at the end of the day I mean it is better we're going to test it out and see but I mean a lot of people I know are going to say hey I know it's cheaper and it is less limited

But that still isn't compared to deep seeks 100% open 100% essentially free model where you're really not limited to and just quickly taking a look at the Deep seek R1 benchmarks as you can see here here are some of them where deep seek is surpassing 01 mini as well as 01 right here in many different tests such as code forces math Etc um with some really crazy scores and like I mentioned on this channel you can't always trust benchmarks especially when

The company making the model is putting the benchmarks out for themselves but let's take a look at some of 03 Mini's benchmarks so we can see on competitive math right here it's going against 01 preview 01 01 mini and then 03 mini low 03 mini medium and 03 mini high and if you're wondering what's the difference between low medium high this essentially the more power from model that's going to run longer think longer with the whole test time compute Paradigm so

Low you're going to get quicker responses but it may not be as quality as high now if we're looking at low 03 mini still beats 01 preview it's just under 01 mini right here and it's a decent amount below 01 now 03 mini medium is above 01 mini and it's only a little bit below 01 and then 03 mini high is above all these models including 01 now you will notice that they aren't showing 01 Pro which I don't think 03

Mini here currently beats and you can see for PhD science level questions we have a more even playing field right here 03 mini is beating 01 mini and it's not beating 01 or 01 preview and then 03 mini medium is just below these two and then 03 mini high is actually number one competition code we can see that it is apparently supposed to write pretty good code comparative to 01 you can see it's scoring around 200 points more uh than 01

And a lot more than 01 mini and actually all three of the 03s are pretty much at the top 03 mini low is just a little bit below 01 software engineering Benchmark is pretty similar we have 03 mini up here above 01 and this is for the high one and then medium is just below 01 and then low is just below 01 preview we can see a latency comparison here between 01 mini and 03 mini medium so 03 mini medium is

Supposed to be a decent amount faster than 01 mini we're seeing 7,500 milliseconds right here for 03 mini and then we're seeing just over 10,000 for 01 mini they have some safety benchmarks like disallowed content evaluations which all these models including deep seek you know are definitely censored in many different ways and then jailbreak evaluations right here so 03 Mini model delivers respones 24% faster than its predecessor 01 mini while making 39% fewer mistakes major mistakes on complex problems it maintains

The ability to show its problem solving process rather than just providing final answers developer features a support so it supports function calling structured outputs streaming capabilities and developer messages now I know a lot of you are wondering how this compares to deep SE R1 and we're going to do a few tests in just a second but just a quick rundown from perplexity here we can see we can see deep seek R1 shows stronger performance in specialized tasks like 79.8% on AIM

Math benchmark versus 03 minis 78.2% it outperforms in complex reasoning tasks and code forces so 20 2029 versus 1997 and then 03 mini performs better in structured dialogue and shows more consistent performance in routine tasks now deep seek R1 is substantially more affordable and we'll get into that in just a second and the contact window for 03 mini is 200k tokens while deep seek has only 128k tokens and the max output is 100k tokens on 03 mini and 32k tokens on

Deep seek R1 03 mini open source no and deep seek open source yes apparently it demonstrates significantly better safety metrics as well and like I mentioned deep seek is substantially cheaper so if we look at Deep SE R1 API we have input API price right here so 0.14 cents per million hit uh cash Miss is 55 cents per million tokens output price API 2.19 million tokens and then if we go ahead and look at the pricing for 03 mini we can

See that it is the exact same as 01 mini so for input it is $110 per million tokens and then the cash input is 0.55 cents per million tokens and then the output is $440 per million tokens so before I show you actually how to use 03 I just want to go over some of the things that people are saying on Twitter so uh McKay Wrigley says I replaced every AI agent and workflow I have running on o Open AI 01

Model with the no new 03 mini they all work in some better but nine times cheaper and four times faster they are significantly under hyping this model it's absolutely unbelievable 03 and 03 Pro will be crazy and I can definitely say that I agree with 03 and 03 Pro will honestly feel like you're maybe in a psychedelic trip when you're talking to it just how intelligent these models will be and it's exciting but also kind of scary to say to see

This now one other thing I want to note is that wind surf right here actually already added 03 mini in Wind Surf and they added this 5 hours ago and open AI posted their post which I'll show you in just a second 5 hours ago so cud to win win Sur for doing that I just made a video yesterday how they just integrated deep seek R1 into their platform so as you can see 03 mini medium reasoning right it cost one

Credit I don't know why they gave it the same credit price as GPT 40 when GPT 40 is over twice the price but anyways here this person says 03 mini might be the best llm for real world physics prompt write me a python script of a bouncing ball inside a test RCT and boom you can see that test RCT right there the bouncing ball this person says 03 mini high does this challenge first time no other model has made working shaders

In many tries let alone one prom is create a visually interesting Shader that can run in twiggle dot app make it like the ocean in a storm make it even more interesting scary good model all right and here's some of open ai's post right here 03 mini is now available in chat GPT and the API Pro users will have unlimited access Plus team users will have triple rate limits and then free users can go ahead and try it out Dam Alman

Tweeted out about this and then this guy said here's what happened deep seek forced some emergency meetings in open Ai and they came up with the solution to keep up the 03 hype I'm glad we're forcing them to move fast and I definitely agree if it wasn't for deep seek I don't know if we'd be getting this model right now so it's good to see that there's some competition in this space and that it's open source all right so to get

Started with 03 mini you're going to go to chat gp.com you're going to go to your model selector and you'll see that you have 03 mini which is fast at Advanced reasoning and then 03 mini High which is great at coding and logic so if you're going to try to be solving very challenging problems you may want to use 03 mini high or if you're coding you may want to use this and then O3 mini if you just need quick reasoning

About certain things so we'll test out all these as well as compare the result with deeps R1 so when you're using 03 mini you can toggle 03 mini with search which is a new feature and of course deep seek has this so opena I had to get it too let's go ahead and keep it off for now and we're going to paste in this prompt right here I'm going to start off with 03 mini the fast version and we're saying create

A one-page website for my Ecom Beauty site the color should be pink white and black it should have light and dark mode it should be very modern and sleek and look very impressive showcase your coding capabilities and add cool unique features and elements to impress me all right we're going to go ahead and send this at the same time while 03 mini is going to work okay that was quick I'm not going to lie that was very quick but it did

Only you know we can see it only had a few different thoughts right here and then it generated the code and I'm also going to run this in deep seek R1 all right so if I go ahead and run this you can see deep seek is definitely a bit slower compared to 03 mini the fast version and while deep seek is generating the code right here let's go ahead and test 03 minis okay so I'm going to go ahead and grab

This 03 mini code and this is actually a lot right here that it generated let's go ahead and just view this okay and we got our beauty store right here so Unleash Your Beauty best sellers this is actually pretty good now this is a very small site let's go ahead and toggle dark mode and boom it actually did a really good job especially for this being the um smaller version now I'm curious to see deep seek so let's crab deep seek

Right here and go to this viewer paste this in okay not bad okay okay it even gave us a little mouse icon here so they both did pretty good I'm going to go ahead and say one more prompt here add more sections in the page make the code longer and have more advanced stuff and while that's loading with deep seek I'm going to do the same thing with 03 mini now O3 mini is thinking very quickly like it only has one

Thought bullet point right here so test time compute for this um model is definitely a lot faster I'm curious to see the 03 mini High version are you tired of pouring thousands of dollars into appointment Setters only to watch leads slip away imagine having a team of elite sales agents booking qualified appointments for you around the clock no more wasted time on training no more frustration with performance and no more draining your budget on inconsistent and expensive call centers introducing stride

Agents AI powered appointment centers that work 24/7 never get tired and book appointments while you sleep trained on thousands of successful conversations are AI agents outperformed human teams at just one tenth of the cost join the ranks of businesses that doubled their appointments and booking rates in just a matter of weeks don't get left behind in the AI Revolution visit stride agents.com now and transform your entire sales process with Cutting Edge AI technology it's time to accelerate your stride with AI

Agents okay this generated a lot of code um yeah a lot so I'm okay let's go ahead and try this new 03 Mini right here in the viewer okay it generated 683 lines of code um as you can see if we're scrolling we got our best sellers right here the lip dick makeup we got an about section we got our features right here uh customer testimonials subscribe to our newsletter and the dark mode is working except on the featur section there's

A little bit and the subscribe to newsletter a little bit of issues there but all in all pretty impressive and of course guys this is just a quick comparison since I'm going to definitely be doing more videos on O3 mini because I'm already quite impressed just by um you know using reasoning models these last few weeks and just in general like I can see that O3 mini is definitely a Powerhouse and deeps seek we got a new generation right here and

It says it's enhanced with some more features I'm curious to see it is a lot less lines of code but let's go ahead and copy this okay so I don't know why I'm getting this I'm sending this back to deep seek and saying hey I pasted this in why is it all white background and black text with no styling worse than the first one okay it says because it gave us the incomplete code okay that makes sense now the one cool

Thing about deep seek I forgot to mention is we could also run the HTML right here which is really nice okay so we're going to add more advanced Styles right here okay nice looking better and then where it says add more advanced interactions for our JS I'm going to paste that in okay so now we actually have a scrolling bar at the top top which is really nice asking is this correct okay so we need to add our nav content right

Here our hero section our custom cursor styling okay so here it is I mean it's you know pretty good uh I think just because of the contacts and whatnot and the generation output it was a little bit funky but I did kind of patch it together um so did a pretty good job now in terms of actually the UI interface if you're using it in the chat with deeps O3 mini you're going to be able to generate rate a lot more

As you can see here okay so now we can see that it is taking a lot longer compared to the regular 03 mini and it is actually thinking through the process and has a lot more bullet points throughout this question it's actually still fairly fast though it's actually faster still than deep seek R1 so as you can see it's generating a lot of code right off the bat okay I haven't even ran this yet but I feel like this is going

To be pretty good for the first generation let's go ahead and test it out wow okay this in my opinion is probably the best maybe the best looking one for the first generation like the other two weren't this long for the first generation let's go ahead to toggle it and it's pretty nice that's pretty impressive I'm going to go ahead and copy this second command uh the second prompt that we ran for these other two and let's see what it actually

Does now add more sections in the page make the code longer and have more advanced features and stuff okay so it's reasoning now enriching the web page crafting a vibrant web page enhancing the web page sections interactivity designing new sections implementing features updating navigation reworking section layout of course this isn't the direct Chain of Thought that you know the model is having like we see in deep seek but it is still pretty cool to see that now let's see what this

Is doing so this is actually lightning fast I'm not going to lie this is very fast still for 03 high and it is generating a lot of code right now we even have a team section a newsletter our footer now it's on the JavaScript okay let's go ahead and test this out I have a feeling this will be the best yet okay and here is our site it is 780 lines of code we can see this is light mode right here

About section what our customers are saying as we scrolling um you know we have animations coming through to show the different elements newsletter contact us we have a scroll to top button and we have our dark mode working with all the different products we have a slider right here for the testimonials accordion right here so definitely I would say in terms of this little small test obviously this isn't an in-depth test this is just a first look testing out some simple

Prompts uh 03 mini large is very impressive in my opinion just from seeing this I'm definitely going to be using this a lot 03 mini high and I'm curious to see when we actually bring it into some of our AI coding agents which I'll do videos on very soon definitely this week make sure to stay tuned for that how it's actually going to perform like when it's doing multifile editing and doing all these different things how is it going to do

But that's pretty much it for this video guys I just wanted to give you a quick update lightning fast as I saw 03 mini come out I actually actually recorded another video earlier today but I'm probably going to post this one first because 03 mini just dropped so other than that guys if you want to stay up to date with all the latest AI news AI updates and how to actually Implement AI into your life and your business to accelerate your

Stride make sure to like the video comment down below and subscribe to stay up to date with all the videos also too guys if you're new here and you haven't already joined our free Facebook group and Discord Channel stric community.com I'll leave a link down below so you can definitely check that out and then also too guys if you haven't already join our stride AI Academy you're definitely going to want to check that out behind the scenes resources courses Etc it's

Going to be jam-packed with loads of free stuff and currently it is free but in the future it may not be so join while you still can link down below and then also too guys if you run a business and you need help implementing AI agents like Ai call center Ai call calling into your business to book boatloads of qualified leads onto your calendar then book a call down below at executive.com apply we help businesses Implement AI agents to accelerate their

Stride let me know what your thoughts are down below about 03 mini does it beat deep seek R1 does it beat China's new models these open source models or is open AI behind I think 03 mini high is extremely powerful and I can't wait to test it out on some other tools other than that guys I will see you in the next video keep hustling keep grinding and of course guys accelerate your stride take care