00:00So the biggest AI image model that we've seen thus far dropped yesterday and it's called
00:04ChatGPT Image 2.0. It by far trumps anything that we've seen to this point. So on the image arena
00:13you can see GPT Image 2 is above Nano Banana 2 which was previously the throne holder if you will
00:23and Image 2 can also use this thinking mode where it can go even deeper into the process of
00:32building out whatever kind of image that you want. So this number at 1512 which is above 1271
00:41which is how they're denominating the quality of the model it can actually go further than that
00:47with the thinking mode. And so you can pretty much do anything with this model and I've seen it for
00:53all sorts of different use cases from product images to ad creative to visualizations graphs dashboards
01:03mock-ups designs and I think what is most impressive about it is that it is now flawless
01:13to generate images with accurate and precise text and that's what I probably like most about it.
01:20So what I thought I would do in this video is actually give a few prompts of how it really
01:26compares in in reality to Nano Banana 2 which again was the previous throne holder. So I've got a few
01:35prompts prepared. I have Nano Banana 2 opened up here and then we have the new GPT Image Model
01:42opened up here. And so the first prompt I'm going to do is around food. So I have here an
01:50overhead
01:50flat lay of a hearty breakfast on a reclaimed wood table, two sunny side up eggs on sourdough toast,
01:59crispy bacon roast, did cherry tomatoes on the vine, a small ramekin of hot sauce, a black ceramic mug of
02:06coffee with steam rising and a linen napkin in the upper right corner, natural morning light streaming
02:11from the left soft shadows styled but lived in shot on the 35 millimeter lens, warm color grade.
02:17So we're going to use Nano Banana 2 for this and then we're also going to put the same prompt
02:21into the new image model and we can use the thinking mode for it which gives you a little bit
02:29different
02:30of a feel and you can also see what it looks like when it has the thinking a part of
02:36it versus when it
02:36doesn't have a part of it as well. So you see how it came out with Nano Banana 2 and
02:42again Nano Banana 2
02:43was also exceptionally good for images and I've pretty much been using it for everything that I
02:49do. Like I use it for one of the image series that I'm doing with my mobile app still app
02:55for I'm doing
02:57this basically like 90 day image series on meditation. My mobile app is a personalized
03:04meditation app and I basically created the series with this mascot. It's a blue mascot that kind of
03:11takes you through different meditation exercises and educates you on meditation. So Nano Banana 2 is
03:18really good. It's really great for thumbnails. It's been really great for all sorts of use cases.
03:21So we're at a point now where to even be elevating beyond where we currently are is pretty impressive
03:30but you can see here the differences in these two images. So this is the Nano Banana 2 example. This
03:36is
03:36the GPT Image 2 example. Like this looks real. Like this looks like it was actually taken at the
03:47table and again they're both good but you can see that this is on another level. You can see it
03:54without the thinking mode. You can see I would say the difference is that with the thinking mode there's
04:01a little bit more depth to it. It just gives more of like a live hyper-realistic view. This one
04:08looks
04:08closer to the Nano Banana example here. So we're just going to keep the thinking mode on for this. I'm
04:14going to move on to the next prompt. So the next prompt is going to be a movie poster. So
04:19this one is
04:19a cinematic movie poster for a fictional thriller titled The Last Signal. A lone figure in a trench
04:23coat stands on a rain-soaked city rooftop at night backlit by a massive growing radio tower.
04:29And so we're going to see what that gives us. All right so here's what we got for Nano Banana
04:332.
04:34So you can see that the text was accurate as Nano Banana typically does anyway. But yeah really solid
04:42kind of movie poster and we'll see how that compares with what GPT Image 2 does. I think
04:48what's also cool about the GPT Image is that it kind of describes what it's doing as it's thinking and
04:54how it's sort of iterating on the design in real time. That's pretty impressive. You can see on the
05:02right side it shows how it's thinking and iterating on the process and how long it's also taking in that
05:09process. And I'll say that the GPT Image Models historically have taken a little bit longer in
05:14order to render. And so it's interesting seeing how it is how the user experience for it is actually
05:22I'd say enhanced somewhat. So in this case it's actually bringing us back is bringing us back four
05:28images. Okay let's bring us back one image. So this is the example that they gave here versus the
05:32example with the Nano Banana. They're both really good. But yeah I'd say that OpenAI version is a
05:40little better. All right next prompt. All right so this next one's going to be a science example
05:44have here. A biology example. So a clean educational cross-section diagram of a human red blood cell and
05:53it's surrounding inside a capillary. So kind of going into cell biology. Let's see what it does here.
05:59Let's see what the GPT Image does as well. Let's see what that brings us back. All right so this
06:04is
06:04what Nano Banana did right here. Red blood cell shows the different parts to it. We got the white
06:10blood cell here. Hemoglobin and the like. And this is the GPT version. A little bit more defined.
06:17Say more along the lines of what I'd want to see in an image like this. I'm going to do
06:21one more
06:21example. So this one is going to be a trading dashboard. We'll see how it compares. Pop that into
06:28Nano Banana. Then into GPT. Let's see what it brings back. All right so one thing about the
06:34Nano Banana examples are Nano Banana is super quick. It's super quick relative to the GPT Image. GPT
06:42Image just takes longer. Granted the thinking mode probably takes more time as well but we have
06:47NQ here. Trading dashboard which looks pretty standard. We'll see what GPT comes back with.
06:54All right so this is GPT version here. I'd say it's pretty similar. It's pretty similar.
07:02Not necessarily sure if I actually prefer one or the other. So yeah this is the new GPT Image 2
07:09.0. This
07:09is how it compares to Nano Banana 2. And I'll say that now have both these models right to use
07:16as
07:17options. And if you look at the benchmarks however GPT Image 2 significantly surpasses everything else.
07:25So this is unequivocally the top model that we now have out. And that is something just kind of keep
07:32in mind. I believe Image 2.0 is 0.006 cents per image is what it comes out to versus
07:41Nano Banana
07:42and it depends on how you're using it. This is definitely another leap forward in the image
07:50models. And I would also say that when you combine something like this with Claude for example launching
07:55recently their Claude design feature you can pretty much do anything that you're that you imagine from a
08:04design standpoint. So something to play around with. Hope this is helpful and that you now understand the
08:11capabilities of GPT Image 2.0. That's all I got for you for this one. And I will see you
08:17guys on the next one.
Comments