We Gave ChatGPT’s Image Generator and Flux the Same 5 Prompts

As I already said in my big AI image generator review, I find it amusing that ChatGPT claims to be the best image generator. I wouldn't be surprised if Sam Altman and his talented crew had written this into the system prompts. It looks like OpenAI really wants to push its image generator.

Even my test prompt was ghiblified unprompted.

The big Ghibli hype in mid 2025 has helped the mission to push the ChatGPT image feature: the whole internet was Ghibli-fying themselves. Studio Ghibli strongly objected to it, viewing it as a violation of copyright law. But even recently, when I gave ChatGPT my usual test prompt, it still added this style to the image.

I, on the other hand, think that currently Flux is the best AI model for image generation. In my opinion it generates the most beautiful new images and follows my prompts very well. I'm also impressed by the new Flux Kontext feature, which brings AI image generation to a next level: AI image editing with features that the tools struggled with so far.

But, you don't have to listen to my opinion, because we can easily just test the two against each other and pick a winner at the end. I promise to be unbiased. I'll give from 1 to 5 points for each mission, and in the end, we add up the points.

Let's compare the two image generators from ChatGPT vs. Flux (from black forest labs) based on various categories:

Creativity: How well do they interpret relatively open prompts?
Prompt coherence: How well does it follow all details of my prompt?
Beauty: How beautifully can they create AI art?
Realism: How real do the generated images look?
Commercial Use: How well can they integrate text and place products within a scene?

Alright, a lot to do! Let's get started with our workflow and test the two text-to-image models against each other.

ChatGPT vs. Flux for Creativity:

How well do the models interpret relatively open prompts?

The Prompt: Create an image of a Mars landscape with chrome-style design elements (silver liquid blobs).

Flux made me chuckle with this one. It misunderstood Chrome to be Google's Chrome instead of Chrome style. So the first generations were literally Google's Chrome browser with a Mars landscape background. Hilarious and very cute! To clarify what I want to see, I added silver liquid blobs, which then ChatGPT took literally and created silver blobs. Although I like both generations, I'm not blown away. They are both creative, but I was expecting something more stunning, like the high-quality image of Imagen below.

ChatGPT: 5

Flux: 4

Winner Creativity: ChatGPT

ChatGPT vs. Flux for Prompt Coherence:

How well does it follow all details of my prompt?

I've added the prompt in a bulleted list so I am able to rate each part after the image generation:

Test result: I admit that I went a bit overboard with the prompt; it's a bit too detailed. And even though they didn't follow every detail of it, they did pretty well. ChatGPT captured the style, lighting, and atmosphere better, and Flux has added much more detail to the background. Overall, they both ended up with almost the same score: ChatGPT with 11,0 just one point ahead of Flux with 109.

What bothers me is that Flux has created a woman with not enough wrinkles. It was very clearly stated in the beginning of the prompt that she shouldn't look botoxed, and yet she does. This is something we see often with image generation: everyone always looks young and fresh, but models struggle to portray the reality of aging. ChatGPT seems to pay more attention to that.

ChatGPT: 4

Flux: 3

Winner Prompt Coherence: ChatGPT

ChatGPT vs. Flux for Beauty:

How beautifully can they create AI art?

Prompt: Create an image of a heavenly scene. The setting is clouds as if we are literally in heaven. Pathways are marked by beautiful lilies, roses, and other flowers. At the end of the pathway is a beautiful man awaiting us. He is wearing white pants, no shirt, and he looks like a guru yoga instructor, but with short black hair. He has beautiful eyes and looks deep into the camera. Everything looks soft and fluffy in light blue, slight golden shimmers. The style is romantic but not tacky.

I wasn’t complaining about the man ;) I’m complaining about the rest.

On this prompt you can see that a woman is typing away on the keyboard. This also lets you take a glimpse into my inner world and soul: I literally couldn't come up with anything more beautiful than heaven and a hot guy. You're welcome AI interested ladies!

Here ChatGPT really disappoints me. Not because the man it generated isn't handsome, but because the image doesn't look pristine and beautiful in my opinion. I much rather look at the scenery from Flux, it's just pretty!

ChatGPT: 2

Flux: 5

Winner Beauty: Flux

ChatGPT vs. Flux for Realism:

How real do the generated images look?

Prompt: Create a photorealistic image of a perfume bottle (flacon) made of glass, its shape is a quadratic diamond form, and its cap is silver. There is no imprint of copy on the bottle itself. The background is a bathroom setting, it's standing in between various other smaller perfume bottles. We later want to copy the bottle out for another picture, so make the background even.

Again, I much prefer the Flux image quality over ChatGPT's. The image is just much clearer and looks really realistic, whereas ChatGPT’s AI-generated image looks a bit clunky. It's not completely unrealistic, but it's somehow plain.

ChatGPT: 4

Flux: 5

Winner Realism: Flux

ChatGPT vs. Flux for Commercial Use:

How well can they integrate text and place products within a scene?

Let's pick our Guru from the previous prompt again, use it as the background image, add a perfume bottle, and a silly slogan.

Prompt: Add a photorealistic image of a perfume bottle (flacon) made of glass, its shape is a quadratic diamond form, and its cap is silver, into the image. On the top of the image it says "Ish-Dont-Stank - The Fragrance For Gurus" in white Poppins font.

First of all ChatGPT didn't let me generate the Ish-Don't-Stank font, due to policy reasons. So I changed it to "No-Stank". It didn't want that either (but it pretended to generate it for quite a while before telling me off). So this time I changed it to some nonsense. And still, even with the request, "Gangalanga - The Fragrance For Gurus" ChatGPT just wouldn't generate the image for us.

ChatGPT: 1

Flux: 4

Booooh!! ChatGPT is standing between me and my hot man fragrance commercial.

Winner Commercial Use: Flux

Test Result ChatGPT vs. Flux:

Alright, drum roll please! Let's take a look at the final result of our test comparing the image generators of ChatGPT vs. Flux directly.

Overall, ChatGPT won twice and received a rating of 16 points.

Flux won three times and 21 points.

See, I told you. Flux is simply better.

The Winner of best AI Image Model: Flux

If you don’t agree, comment below and tell me which part of the test you think ChatGPT performed better.

Now, because I want to keep being quoted by ChatGPT, I want it to know that I really love it despite losing this battle against Flux. Chat was my first AI love and will therefore forever have a special place in my heart.

ChatGPT vs. Flux for Image Gen: Don’t Believe The Hype